File: | platform/mac/avmshell/../../../core/StringObject.cpp |
Location: | line 1257, column 13 |
Description: | Value stored to 'charsUsed' is never read |
1 | /* -*- Mode: C++; c-basic-offset: 4; indent-tabs-mode: nil; tab-width: 4 -*- */ |
2 | /* vi: set ts=4 sw=4 expandtab: (add to ~/.vimrc: set modeline modelines=5) */ |
3 | /* ***** BEGIN LICENSE BLOCK ***** |
4 | * Version: MPL 1.1/GPL 2.0/LGPL 2.1 |
5 | * |
6 | * The contents of this file are subject to the Mozilla Public License Version |
7 | * 1.1 (the "License"); you may not use this file except in compliance with |
8 | * the License. You may obtain a copy of the License at |
9 | * http://www.mozilla.org/MPL/ |
10 | * |
11 | * Software distributed under the License is distributed on an "AS IS" basis, |
12 | * WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License |
13 | * for the specific language governing rights and limitations under the |
14 | * License. |
15 | * |
16 | * The Original Code is[Open Source Virtual Machine.]. |
17 | * |
18 | * The Initial Developer of the Original Code is |
19 | * Adobe System Incorporated. |
20 | * Portions created by the Initial Developer are Copyright (C) 2008 |
21 | * the Initial Developer. All Rights Reserved. |
22 | * |
23 | * Contributor(s): |
24 | * Michel Daumling |
25 | * |
26 | * Alternatively, the contents of this file may be used under the terms of |
27 | * either the GNU General Public License Version 2 or later (the "GPL"), or |
28 | * the GNU Lesser General Public License Version 2.1 or later (the "LGPL"), |
29 | * in which case the provisions of the GPL or the LGPL are applicable instead |
30 | * of those above. If you wish to allow use of your version of this file only |
31 | * under the terms of either the GPL or the LGPL, and not to allow others to |
32 | * use your version of this file under the terms of the MPL, indicate your |
33 | * decision by deleting the provisions above and replace them with the notice |
34 | * and other provisions required by the GPL or the LGPL. If you do not delete |
35 | * the provisions above, a recipient may use your version of this file under |
36 | * the terms of any one of the MPL, the GPL or the LGPL. |
37 | * |
38 | * ***** END LICENSE BLOCK ***** */ |
39 | |
40 | #include "avmplus.h" |
41 | |
42 | // This is the maximum value for the characters Left field in the |
43 | // m_bitsAndFlags field of the String instance. Strings grow by |
44 | // by doubling the available buffer size until the characters left |
45 | // exceeds the size of the field. From there on, the buffer grows |
46 | // linear. Note that this number is divided by two to make sure that |
47 | // the counter actually, spans the entire buffer area; if this would |
48 | // not be done, the calculation of characters left (using GC::Size()) |
49 | // would return a wrong number. |
50 | #define TSTR_MAX_CHARSLEFT((uint32_t)TSTR_CHARSLEFT_MASK >> TSTR_CHARSLEFT_SHIFT >> 1) ((uint32_t)TSTR_CHARSLEFT_MASK >> TSTR_CHARSLEFT_SHIFT >> 1) |
51 | |
52 | // This is the maximum value of extra bytes to allocate during a |
53 | // string concatenation operation if PleaseAlloc() fails. The number |
54 | // is quite small to not kill the app just because of a large extra |
55 | // chunk of RAM being allocated just in case there are more concatenation |
56 | // operations. |
57 | #define TSTR_MAX_LOMEM_EXTRABYTES16384 16384 |
58 | |
59 | // Never allocate dynamic strings smaller than this (in bytes) |
60 | #define TSTR_MIN_DYNAMIC_ALLOCATION32 32 |
61 | |
62 | // in fixDependentString, don't bother going dependent->dynamic if the memory |
63 | // we would save is < this (in bytes) |
64 | #define TSTR_DEPENDENT_STRING_NUISANCE_SAVINGSint32_t(sizeof(String)) int32_t(sizeof(String)) |
65 | |
66 | using namespace MMgc; |
67 | |
68 | namespace avmplus |
69 | { |
70 | // Use this constant for 16 and 32 bit strings of zero length |
71 | // we do not want any string to contain NULL buffer pointers |
72 | union Zero |
73 | { |
74 | uint32_t u32; |
75 | wchar u16; |
76 | uint8_t u8; |
77 | char c8; |
78 | }; |
79 | static const Zero k_zero = { 0 }; |
80 | |
81 | /////////////////////////// Helpers: Widening ////////////////////////////// |
82 | |
83 | REALLY_INLINEinline __attribute__((always_inline)) void _widen8_16(const uint8_t* src, wchar* dst, int32_t len) |
84 | { |
85 | while (len-- > 0) |
86 | *dst++ = wchar(*src++); |
87 | } |
88 | |
89 | /////////////////////////// Helpers: Narrowing ////////////////////////////// |
90 | |
91 | REALLY_INLINEinline __attribute__((always_inline)) void _narrow16_8(const wchar* src, uint8_t* dst, int32_t len) |
92 | { |
93 | while (len-- > 0) |
94 | { |
95 | AvmAssert(*src <= 0xFF)do { } while (0); |
96 | *dst++ = (uint8_t) *src++; |
97 | } |
98 | } |
99 | |
100 | /////////////////////////// Helpers: Copying /////////////////////////////// |
101 | |
102 | // Copy the contents of the given buffer to another buffer, given two widths. |
103 | |
104 | static void* _copyBuffers(const void* src, void* dst, int32_t srcLen, String::Width srcWidth, String::Width dstWidth) |
105 | { |
106 | if (srcWidth == dstWidth) |
107 | VMPI_memcpy::memcpy(dst, src, srcLen << srcWidth); |
108 | else if (srcWidth == String::k8) |
109 | _widen8_16((const uint8_t*) src, (wchar*) dst, srcLen); |
110 | else |
111 | _narrow16_8((const wchar*) src, (uint8_t*) dst, srcLen); |
112 | // return the new buffer pointer |
113 | return (char*) dst + (srcLen << dstWidth); |
114 | } |
115 | |
116 | #ifdef _DEBUG |
117 | void String::verify7bit() const |
118 | { |
119 | if (!(m_bitsAndFlags & TSTR_7BIT_FLAG)) |
120 | return; |
121 | |
122 | if (getWidth() == k8) |
123 | { |
124 | Pointers ptrs(this); |
125 | for (int32_t i = 0; i < m_length; ++i) |
126 | { |
127 | if (ptrs.p8[i] > 127) |
128 | { |
129 | AvmAssert(!"hi bit found on string with TSTR_7BIT_FLAG set")do { } while (0); |
130 | } |
131 | } |
132 | } |
133 | else |
134 | { |
135 | AvmAssert(!"only k8 strings should set TSTR_7BIT_FLAG")do { } while (0); |
136 | } |
137 | } |
138 | #endif |
139 | |
140 | /////////////////////////// Helper: get AvmCore ///////////////////////////// |
141 | |
142 | REALLY_INLINEinline __attribute__((always_inline)) MMgc::GC* _gc(const String* s) |
143 | { |
144 | return MMgc::GC::GetGC(s); |
145 | } |
146 | |
147 | REALLY_INLINEinline __attribute__((always_inline)) AvmCore* _core(const String* s) |
148 | { |
149 | MMgc::GC *gc = MMgc::GC::GetGC(s); |
150 | return gc->core(); |
151 | } |
152 | |
153 | /////////////////////////// Helpers: templated functions ///////////////////////////// |
154 | |
155 | // |
156 | // use typetraits to generate a compile-time error if you attempt to use a |
157 | // non-unsigned char type as an argument to these. |
158 | // |
159 | // Note that we deliberately are using a bitwise &, not a logical && here... |
160 | // GCC 4.3 and later, with -Wlogical-op enabled, will complain about the && |
161 | // construct with |
162 | // |
163 | // "warning: logical '&&' with non-zero constant will always evaluate as true" |
164 | // |
165 | // and since many embedders compile with warnings-as-errors enabled, this crushes |
166 | // the build. This seems like a highly questionable warning to me, but since there |
167 | // is a valid workaround, we'll go with it rather than requiring embedders to disable |
168 | // that warning. |
169 | // |
170 | #define PREVENT_SIGNED_CHAR_PTR(TYPE){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, TYPE> is_char; typedef :: MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_170; } \ |
171 | { \ |
172 | typedef MMgc::is_same<char, uint8_t> is_char_unsigned; \ |
173 | typedef MMgc::is_same<char, TYPE> is_char; \ |
174 | MMGC_STATIC_ASSERT(!(is_char::value & !is_char_unsigned::value))typedef ::MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_174; \ |
175 | } |
176 | |
177 | // apparently SunPro compiler doesn't like combining REALLY_INLINE with static functions. |
178 | template <typename STR1, typename STR2> |
179 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) boolbool equalsImpl(const STR1* str1, const STR2* str2, int32_t len) |
180 | { |
181 | PREVENT_SIGNED_CHAR_PTR(STR1){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR1> is_char; typedef :: MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_181; } |
182 | PREVENT_SIGNED_CHAR_PTR(STR2){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR2> is_char; typedef :: MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_182; } |
183 | |
184 | for (int32_t j = 0; j < len; j++) |
185 | { |
186 | if (str1[j] != str2[j]) |
187 | return falsefalse; |
188 | } |
189 | return truetrue; |
190 | } |
191 | |
192 | // NOTE: this routine cannot use memcmp(), as the test |
193 | // ecma3/String/localeCompare_rt.as depends on the difference |
194 | // in character values, and memcmp() does not guarantee to |
195 | // return this value |
196 | |
197 | // apparently SunPro compiler doesn't like combining REALLY_INLINE with static functions. |
198 | template <typename STR1, typename STR2> |
199 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t compareImpl(const STR1* str1, const STR2* str2, int32_t len) |
200 | { |
201 | PREVENT_SIGNED_CHAR_PTR(STR1){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR1> is_char; typedef :: MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_201; } |
202 | PREVENT_SIGNED_CHAR_PTR(STR2){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR2> is_char; typedef :: MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_202; } |
203 | |
204 | int32_t res = 0; |
205 | while (len-- > 0 && !res) |
206 | { |
207 | res = int32_t(*str2++ - *str1++); |
208 | } |
209 | return res; |
210 | } |
211 | |
212 | // apparently SunPro compiler doesn't like combining REALLY_INLINE with static functions. |
213 | template <typename STR, typename PATTERN> |
214 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t indexOfImpl(const STR* str, int32_t start, int32_t right, const PATTERN* pat, int32_t patlen) |
215 | { |
216 | PREVENT_SIGNED_CHAR_PTR(STR){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR> is_char; typedef ::MMgc ::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_216; } |
217 | PREVENT_SIGNED_CHAR_PTR(PATTERN){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, PATTERN> is_char; typedef ::MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_217; } |
218 | AvmAssert(patlen > 0)do { } while (0); |
219 | |
220 | // even with REALLY_INLINE, some compilers will be reluctant to inline equalsImpl here, |
221 | // so we explicitly repeat the code here. |
222 | const PATTERN pat0 = pat[0]; |
223 | const STR* const end = str + right; |
224 | for (const STR* probe = str + start; probe <= end; ++probe) |
225 | { |
226 | if (probe[0] == pat0) |
227 | { |
228 | for (int32_t j = 1; j < patlen; j++) |
229 | { |
230 | if (probe[j] != pat[j]) |
231 | goto no_match; |
232 | } |
233 | return int32_t(uintptr_t(probe - str)); |
234 | } |
235 | no_match: |
236 | // some compilers will complain about a label with no statement following, |
237 | // hence the continue |
238 | continue; |
239 | } |
240 | |
241 | return -1; |
242 | } |
243 | |
244 | // apparently SunPro compiler doesn't like combining REALLY_INLINE with static functions. |
245 | template <typename STR> |
246 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t indexOfCharCodeImpl(const STR* str, int32_t start, int32_t right, wchar c) |
247 | { |
248 | PREVENT_SIGNED_CHAR_PTR(STR){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR> is_char; typedef ::MMgc ::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_248; } |
249 | |
250 | const STR* p = str + start - 1; |
251 | const STR* const end = str + right; |
252 | while (++p <= end) |
253 | if (*p == c) |
254 | return int32_t(p - str); |
255 | |
256 | return -1; |
257 | } |
258 | |
259 | // apparently SunPro compiler doesn't like combining REALLY_INLINE with static functions. |
260 | template <typename STR, typename PATTERN> |
261 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t lastIndexOfImpl(const STR* str, int32_t const start, const PATTERN* pat, int32_t patlen) |
262 | { |
263 | PREVENT_SIGNED_CHAR_PTR(STR){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR> is_char; typedef ::MMgc ::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_263; } |
264 | PREVENT_SIGNED_CHAR_PTR(PATTERN){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, PATTERN> is_char; typedef ::MMgc::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_264; } |
265 | AvmAssert(patlen > 0)do { } while (0); |
266 | AvmAssert(start >= 0)do { } while (0); |
267 | |
268 | // even with REALLY_INLINE, some compilers will be reluctant to inline equalsImpl here, |
269 | // so we explicitly repeat the code here. |
270 | const PATTERN pat0 = pat[0]; |
271 | for (const STR* probe = str + start; probe >= str; --probe) |
272 | { |
273 | if (probe[0] == pat0) |
274 | { |
275 | for (int32_t j = 1; j < patlen; j++) |
276 | { |
277 | if (pat[j] != probe[j]) |
278 | goto no_match; |
279 | } |
280 | return int32_t(uintptr_t(probe - str)); |
281 | } |
282 | no_match: |
283 | // some compilers will complain about a label with no statement following, |
284 | // hence the continue |
285 | continue; |
286 | } |
287 | return -1; |
288 | } |
289 | |
290 | // apparently SunPro compiler doesn't like combining REALLY_INLINE with static functions. |
291 | template <typename STR> |
292 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t hashCodeImpl(const STR* str, int32_t len) |
293 | { |
294 | PREVENT_SIGNED_CHAR_PTR(STR){ typedef MMgc::is_same<char, uint8_t> is_char_unsigned ; typedef MMgc::is_same<char, STR> is_char; typedef ::MMgc ::static_assert_MMgc<sizeof (::MMgc::STATIC_ASSERTION_FAILED <(bool)(!(is_char::value & !is_char_unsigned::value))> )> MMgc_static_assert_line_294; } |
295 | |
296 | // must be same signed-ness as other hashcode functions. |
297 | // experimentation shows better results from signed (vs unsigned). |
298 | int32_t hashCode = 0; |
299 | while (len--) |
300 | hashCode = (hashCode >> 28) ^ (hashCode << 4) ^ *str++; |
301 | return hashCode; |
302 | } |
303 | |
304 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t hashCodeAddChar(int32_t hashCode, char val) |
305 | { |
306 | // must be same signed-ness as other hashcode functions. |
307 | // (match sign propagation in right-shift >> 28, for better or worse) |
308 | return ((hashCode >> 28) ^ (hashCode << 4) ^ val); |
309 | } |
310 | |
311 | /*static*/ REALLY_INLINEinline __attribute__((always_inline)) int32_t stringLengthUInt(uint32_t value) |
312 | { |
313 | if (value < 10) return 1; |
314 | else if (value < 100) return 2; |
315 | else if (value < 1000) return 3; |
316 | else if (value < 10000) return 4; |
317 | else if (value < 100000) return 5; |
318 | else if (value < 1000000) return 6; |
319 | else if (value < 10000000) return 7; |
320 | else if (value < 100000000) return 8; |
321 | else if (value < 1000000000) return 9; |
322 | else return 10; |
323 | } |
324 | |
325 | ////////////////////////////// Constructors //////////////////////////////// |
326 | |
327 | // ctor for a static string. |
328 | |
329 | REALLY_INLINEinline __attribute__((always_inline)) String::String(const void* buffer, Width w, int32_t length, boolbool is7bit) : |
330 | #ifdef DEBUGGER |
331 | AvmPlusScriptableObject(sotString()), |
332 | #endif |
333 | m_buffer(buffer), // static data - no WB() needed |
334 | m_extra(NULL__null), |
335 | m_length(length), |
336 | m_bitsAndFlags(w | (kStatic << TSTR_TYPE_SHIFT) | (uint32_t(is7bit) << TSTR_7BIT_SHIFT)) |
337 | { |
338 | AvmAssert(m_length >= 0)do { } while (0); |
339 | AvmAssert((uint64_t(m_length) << getWidth()) <= 0x7FFFFFFFU)do { } while (0); |
340 | } |
341 | |
342 | // ctor for a dynamic string. |
343 | |
344 | REALLY_INLINEinline __attribute__((always_inline)) String::String(MMgc::GC* gc, void* buffer, Width w, int32_t length, int32_t charsLeft, boolbool is7bit) : |
345 | #ifdef DEBUGGER |
346 | AvmPlusScriptableObject(sotString()), |
347 | m_buffer((void*)NULL__null), |
348 | #else |
349 | m_buffer(buffer), |
350 | #endif |
351 | m_extra(NULL__null), |
352 | m_length(length), |
353 | m_bitsAndFlags(w | (kDynamic << TSTR_TYPE_SHIFT) | (charsLeft << TSTR_CHARSLEFT_SHIFT) | (uint32_t(is7bit) << TSTR_7BIT_SHIFT)) |
354 | { |
355 | AvmAssert(m_length >= 0)do { } while (0); |
356 | AvmAssert((uint64_t(m_length) << getWidth()) <= 0x7FFFFFFFU)do { } while (0); |
357 | #ifdef DEBUGGER |
358 | /* |
359 | an explicit WB is only necessary if there's the possibility of an allocation between the time |
360 | the container (String) is allocated and time we set the field. In nondebugger builds, there is no |
361 | such possibility, but in DEBUGGER builds, the superclass ctor (AvmPlusScriptableObject) can |
362 | cause an allocation if sampling is enabled. ("Icky and brittle", but that's how it is.) |
363 | */ |
364 | WB(gc, this, &this->m_buffer.pv, buffer)gc->privateWriteBarrier(this, &this->m_buffer.pv, ( const void *) (buffer)); |
365 | #else |
366 | (void)gc; |
367 | #endif |
368 | } |
369 | |
370 | // ctor for a dependent string. |
371 | |
372 | REALLY_INLINEinline __attribute__((always_inline)) String::String(MMgc::GC* gc, Stringp master, int32_t start, int32_t length) : |
373 | #ifdef DEBUGGER |
374 | AvmPlusScriptableObject(sotString()), |
375 | #endif |
376 | m_buffer(uintptr_t(start << master->getWidth())), |
377 | #ifdef DEBUGGER |
378 | m_extra(NULL__null), |
379 | #else |
380 | m_extra(master), |
381 | #endif |
382 | m_length(length), |
383 | // note that we propagate TSTR_7BIT_FLAG: if the entire master is TSTR_7BIT_FLAG, then so is the dependent string. |
384 | // @todo: a dependent string could qualify for TSTR_7BIT_FLAG even if it's master is not. worth checking for? |
385 | m_bitsAndFlags((master->m_bitsAndFlags & (TSTR_WIDTH_MASK | TSTR_7BIT_FLAG)) | (kDependent << TSTR_TYPE_SHIFT)) |
386 | { |
387 | AvmAssert(m_length >= 0)do { } while (0); |
388 | AvmAssert((uint64_t(m_length) << getWidth()) <= 0x7FFFFFFFU)do { } while (0); |
389 | #ifdef DEBUGGER |
390 | WBRC(gc, this, &this->m_extra.master, master)gc->privateWriteBarrierRC(this, &this->m_extra.master , (const void *) (master)); |
391 | #else |
392 | /* |
393 | an explicit WBRC is only necessary if there's the possibility of an allocation between the time |
394 | the container (String) is allocated and time we set the field. In nondebugger builds, there is no |
395 | such possibility, but in DEBUGGER builds, the superclass ctor (AvmPlusScriptableObject) can |
396 | cause an allocation if sampling is enabled. ("Icky and brittle", but that's how it is.) |
397 | */ |
398 | AvmAssert(master != NULL)do { } while (0); |
399 | master->IncrementRef(); |
400 | (void)gc; |
401 | #endif |
402 | } |
403 | |
404 | // add a and b and check for overflow |
405 | |
406 | static int32_t int32AddChecked(int32_t a, int32_t b) |
407 | { |
408 | if ((a | b) >= 0) // both nonnegative? |
409 | { |
410 | uint64_t x = uint64_t(a) + uint64_t(b); |
411 | if (x <= 0x7FFFFFFFU) |
412 | return int32_t(x); |
413 | } |
414 | GCHeap::SignalObjectTooLarge(); |
415 | /*NOTREACHED*/ |
416 | return 0; |
417 | } |
418 | |
419 | // shift a left by b and check for overflow |
420 | |
421 | static int32_t int32ShlChecked(int32_t a, int32_t b) |
422 | { |
423 | AvmAssert(b < 32)do { } while (0); // ok for this to be DEBUG-only, the shift amounts are known constants for practical purposes |
424 | if ((a | b) >= 0) // both nonnegative? |
425 | { |
426 | uint64_t x = uint64_t(a) << uint64_t(b); |
427 | if (x <= 0x7FFFFFFFU) |
428 | return int32_t(x); |
429 | } |
430 | GCHeap::SignalObjectTooLarge(); |
431 | /*NOTREACHED*/ |
432 | return 0; |
433 | } |
434 | |
435 | static uint32_t uint32ShlChecked(uint32_t a, uint32_t b) |
436 | { |
437 | AvmAssert(b < 32)do { } while (0); // ok for this to be DEBUG-only, the shift amounts are known constants for practical purposes |
438 | uint64_t x = uint64_t(a) << uint64_t(b); |
439 | if (x <= 0xFFFFFFFFU) |
440 | return uint32_t(x); |
441 | GCHeap::SignalObjectTooLarge(); |
442 | /*NOTREACHED*/ |
443 | return 0; |
444 | } |
445 | |
446 | // Private static method to create a dependent string |
447 | |
448 | // in nondebug builds this is a trivial wrapper around the ctor (which is itself trivial), so REALLY_INLINE it |
449 | REALLY_INLINEinline __attribute__((always_inline)) /*static*/ Stringp String::createDependent(GC* gc, Stringp master, int32_t start, int32_t len) |
450 | { |
451 | AvmAssert(len >= 0)do { } while (0); |
452 | // master cannot be a dependent string |
453 | AvmAssert(!master->isDependent())do { } while (0); |
454 | MMGC_MEM_TAG( "Strings" ); |
455 | Stringp s = new(gc, MMgc::kExact) String(gc, master, start, len); |
456 | VERIFY_7BIT(s)do { } while (0); |
457 | return s; |
458 | } |
459 | |
460 | // Private static method to create a dynamic string, given a buffer and its size in characters |
461 | |
462 | /*static*/ Stringp String::createDynamic(GC* gc, const void* data, int32_t len, Width w, boolbool is7bit, int32_t extra) |
463 | { |
464 | AvmAssert(w != kAuto)do { } while (0); |
465 | AvmAssert(len >= 0)do { } while (0); |
466 | |
467 | // only 8-bit strings should set the 7-bit flag |
468 | if (w != k8) |
469 | is7bit = falsefalse; |
470 | |
471 | // a zero-length dynamic string is legal, but a zero-length GC allocation is not. |
472 | int32_t alloc = int32AddChecked(len, extra); |
473 | |
474 | MMGC_MEM_TAG( "Strings" ); |
475 | |
476 | // First, use PleaseAlloc(), and if the call fails, reduce the amount of extra data |
477 | // to TSTR_MAX_EXTRA_BYTES_IN_LOW_MEMORY and do an Alloc(), which may fail. |
478 | void* buffer = gc->PleaseAlloc(int32ShlChecked(alloc, w), 0); |
479 | if (buffer == NULL__null) |
480 | { |
481 | if (extra > (TSTR_MAX_LOMEM_EXTRABYTES16384 >> w)) |
482 | extra = TSTR_MAX_LOMEM_EXTRABYTES16384 >> w; |
483 | alloc = len + extra; // This is safe because the new value of 'extra' is smaller than the old, and the old was checked |
484 | buffer = gc->Alloc(alloc << w, 0); // Ditto |
485 | } |
486 | |
487 | int32_t bufLen = (int32_t) (GC::Size(buffer) >> w); // Note bufLen may be larger than (alloc << w) |
488 | int32_t charsLeft = bufLen - len; |
489 | // the extra character must not exceed the available field size. |
490 | // |
491 | // ok for this to be an assertion because we will not have problems with mmgc rounding up |
492 | // to a 4KB boundary; the max chars left is about 4KB. |
493 | AvmAssert(charsLeft <= int32_t((uint32_t) TSTR_CHARSLEFT_MASK >> TSTR_CHARSLEFT_SHIFT))do { } while (0); |
494 | |
495 | Stringp s = new(gc, MMgc::kExact) String(gc, buffer, w, len, charsLeft, is7bit); |
496 | |
497 | if (data != NULL__null && len != 0) |
498 | VMPI_memcpy::memcpy(buffer, data, size_t(len << w)); // This is safe because alloc >= len and buffer size is alloc << w and that has been checked. |
499 | #ifdef _DEBUG |
500 | // Terminate string with 0 for better debugging display |
501 | if (charsLeft) |
502 | { |
503 | if (w == k8) |
504 | s->m_buffer.p8[len] = 0; |
505 | else |
506 | s->m_buffer.p16[len] = 0; |
507 | } |
508 | if (data != NULL__null) |
509 | { |
510 | // don't try to verify if data==null, buffer will be random! (caller will verify) |
511 | VERIFY_7BIT(s)do { } while (0); |
512 | } |
513 | #endif |
514 | return s; |
515 | } |
516 | |
517 | // Private static method to create a string, given a static buffer and its size in characters |
518 | |
519 | /*static*/ Stringp String::createStatic(GC* gc, const void* data, int32_t len, Width w, boolbool is7bit) |
520 | { |
521 | AvmAssert(w != kAuto)do { } while (0); |
522 | AvmAssert(len >= 0)do { } while (0); |
523 | MMGC_MEM_TAG( "Strings" ); |
524 | // only 8-bit strings should set the 7-bit flag |
525 | if (w != k8) |
526 | is7bit = falsefalse; |
527 | Stringp s = new (gc, MMgc::kExact) String(data, w, len, is7bit); |
528 | VERIFY_7BIT(s)do { } while (0); |
529 | return s; |
530 | } |
531 | |
532 | // Create a string out of an 8bit buffer. Characters are just widened and copied, not interpreted as UTF8. |
533 | |
534 | Stringp String::createLatin1(AvmCore* core, const char* buffer, int32_t len, Width desiredWidth, boolbool staticBuf) |
535 | { |
536 | if (buffer == NULL__null) |
537 | { |
538 | len = 0; |
539 | buffer = &k_zero.c8; |
540 | staticBuf = truetrue; |
541 | } |
542 | if (len < 0) |
543 | len = Length(buffer); |
544 | |
545 | if (desiredWidth == kAuto) |
546 | desiredWidth = k8; |
547 | |
548 | if (desiredWidth == k8 && core->kEmptyString != NULL__null) |
549 | { |
550 | // core has been initialized, check for cached characters |
551 | if (len == 0) |
552 | return core->kEmptyString; |
553 | |
554 | if (len == 1 && *((uint8_t*) buffer) < 128) |
555 | return core->cachedChars[*((uint8_t*) buffer)]; |
556 | } |
557 | |
558 | const boolbool is7bit = falsefalse; // actually, might be, we just haven't checked yet. |
559 | |
560 | Stringp s = NULL__null; |
561 | GC* gc = core->GetGC(); |
562 | if (staticBuf && desiredWidth == k8) |
563 | { |
564 | s = createStatic(gc, buffer, len, k8, is7bit); |
565 | } |
566 | else |
567 | { |
568 | s = createDynamic(gc, NULL__null, len, desiredWidth, is7bit); |
569 | _copyBuffers(buffer, s->m_buffer.p8, len, k8, desiredWidth); |
570 | VERIFY_7BIT(s)do { } while (0); |
571 | } |
572 | return s; |
573 | } |
574 | |
575 | Stringp String::getFixedWidthString(Width w) const |
576 | { |
577 | if (w == getWidth()) |
578 | return (Stringp) this; |
579 | |
580 | if (w == kAuto) |
581 | return NULL__null; |
582 | |
583 | const boolbool is7bit = falsefalse; |
584 | Stringp newStr = createDynamic(_gc(this), NULL__null, m_length, w, is7bit); |
585 | |
586 | Pointers ptrs(this); |
587 | Pointers new_ptrs(newStr); |
588 | _copyBuffers(ptrs.pv, new_ptrs.pv, m_length, getWidth(), w); |
589 | |
590 | VERIFY_7BIT(newStr)do { } while (0); |
591 | return newStr; |
592 | } |
593 | |
594 | /////////////////////////////// Destructors & tracers //////////////////////////////// |
595 | |
596 | String::~String() |
597 | { |
598 | switch (getType()) |
599 | { |
600 | case kDynamic: { |
601 | // Never necessary to WB() when we store NULL over a non-RC pointer |
602 | void* pv = m_buffer.pv; |
603 | m_buffer.pv = NULL__null; // Avoid dangling the pointer |
604 | _gc(this)->FreeNotNull(pv); |
605 | break; |
606 | } |
607 | case kDependent: |
608 | // WBRC() is however necessary when we store NULL over an RC pointer |
609 | WBRC_NULL(&m_extra.master)MMgc::GC::WriteBarrierRC_dtor(&m_extra.master); |
610 | break; |
611 | default: ; // kStatic |
612 | } |
613 | m_extra.master = NULL__null; // might have already been cleared above, but that's ok |
614 | m_buffer.p8 = NULL__null; // might have already been cleared above, but that's ok |
615 | m_length = 0; |
616 | m_bitsAndFlags = 0; |
617 | } |
618 | |
619 | /*virtual*/ |
620 | boolbool String::gcTrace(GC* gc, size_t cursor) |
621 | { |
622 | (void)cursor; |
623 | switch (getType()) |
624 | { |
625 | case kDynamic: |
626 | gc->TraceLocation(&m_buffer.pv); |
627 | break; |
628 | case kDependent: |
629 | gc->TraceLocation(&m_extra.master); |
630 | break; |
631 | } |
632 | return falsefalse; |
633 | } |
634 | |
635 | /////////////////////////////// Conversions //////////////////////////////// |
636 | |
637 | void String::makeDynamic(const uint8_t* dataStart, uint32_t dataSize) |
638 | { |
639 | AvmAssert(dataStart != NULL)do { } while (0); |
640 | AvmAssert(dataSize > 0)do { } while (0); |
641 | switch(getType()) |
642 | { |
643 | case kStatic: |
644 | // do not convert if range is given and data falls within |
645 | if ((uint32_t) (m_buffer.p8 - dataStart) >= dataSize) |
646 | break; |
647 | // else fall thru |
648 | case kDependent: |
649 | convertToDynamic(); |
650 | break; |
651 | default:; |
652 | } |
653 | } |
654 | |
655 | void String::fixDependentString() |
656 | { |
657 | if (!isDependent()) |
658 | return; |
659 | |
660 | // If the length of this instance is > the master length, this |
661 | // string is the result of an auto-concat; we keep the master in |
662 | // in that case - there is little to no memory gain to be expected. |
663 | // Also, forget about conversion if the memory to be gained is minimal. |
664 | if (m_length >= (m_extra.master->m_length - (TSTR_DEPENDENT_STRING_NUISANCE_SAVINGSint32_t(sizeof(String)) >> getWidth()))) |
665 | return; |
666 | |
667 | // Do not convert to a static string here, because a static string |
668 | // may be converted to a dynamic string, which would leave leave an interior |
669 | // pointer to the static content in the string instance. |
670 | convertToDynamic(); |
671 | } |
672 | |
673 | void String::convertToDynamic() |
674 | { |
675 | AvmAssert(getType() != kDynamic)do { } while (0); |
676 | // Convert this string to be a dynamic string |
677 | int32_t bytes = m_length << getWidth(); // No overflow by definition |
678 | GC* gc = _gc(this); |
679 | MMGC_MEM_TYPE( this ); |
680 | void* buf = gc->Alloc(bytes, 0); |
681 | VMPI_memcpy::memcpy(buf, Pointers(this).pv, bytes); |
682 | WB(gc, this, &this->m_buffer.pv, buf)gc->privateWriteBarrier(this, &this->m_buffer.pv, ( const void *) (buf)); |
683 | if (isDependent()) |
684 | WBRC_NULL(&m_extra.master)MMgc::GC::WriteBarrierRC_dtor(&m_extra.master); |
685 | setType(kDynamic); |
686 | } |
687 | |
688 | /////////////////////////////// Comparison ///////////////////////////////// |
689 | |
690 | int32_t String::Compare(String& other, int32_t other_start, int32_t other_length) const |
691 | { |
692 | if (this == &other) |
693 | return 0; |
694 | |
695 | if (other_start >= other.m_length) |
696 | return -1; |
697 | |
698 | if (other_length < 1 || other_length > other.m_length) |
699 | other_length = other.m_length; |
700 | if (other_start < 0) |
701 | other_start = 0; |
702 | |
703 | int32_t result = 0; |
704 | |
705 | int32_t len = (m_length < other_length) ? m_length : other_length; // choose smaller of two |
706 | if (len > 0) |
707 | { |
708 | Pointers ptrs(this); |
709 | Pointers other_ptrs(&other); |
710 | Width const w1 = this->getWidth(); |
711 | Width const w2 = other.getWidth(); |
712 | switch ((w1 << 1) + w2) |
713 | { |
714 | case (k8 << 1) + k8: |
715 | result = compareImpl(ptrs.p8, other_ptrs.p8 + other_start, len); |
716 | break; |
717 | |
718 | case (k8 << 1) + k16: |
719 | result = compareImpl(ptrs.p8, other_ptrs.p16 + other_start, len); |
720 | break; |
721 | |
722 | case (k16 << 1) + k8: |
723 | result = compareImpl(ptrs.p16, other_ptrs.p8 + other_start, len); |
724 | break; |
725 | |
726 | case (k16 << 1) + k16: |
727 | result = compareImpl(ptrs.p16, other_ptrs.p16 + other_start, len); |
728 | break; |
729 | } |
730 | } |
731 | |
732 | if (result == 0) |
733 | { |
734 | // catch substring compares |
735 | result = (other_length - m_length); |
736 | if (result < 0) |
737 | result = -1; |
738 | else if (result > 0) |
739 | result = 1; |
740 | } |
741 | return result; |
742 | } |
743 | |
744 | boolbool String::equalsLatin1(const char* p, int32_t len) const |
745 | { |
746 | if (len < 0) |
747 | len = Length(p); |
748 | |
749 | if (len != length()) |
750 | return falsefalse; |
751 | |
752 | Pointers ptrs(this); |
753 | if (getWidth() == k8) |
754 | { |
755 | return equalsImpl(ptrs.p8, (const uint8_t*)p, len); |
756 | } |
757 | else |
758 | { |
759 | return equalsImpl(ptrs.p16, (const uint8_t*)p, len); |
760 | } |
761 | } |
762 | |
763 | boolbool String::equalsUTF16(const wchar* p, int32_t len) const |
764 | { |
765 | if (len != length()) |
766 | return falsefalse; |
767 | |
768 | Pointers ptrs(this); |
769 | if (getWidth() == k8) |
770 | { |
771 | return equalsImpl(ptrs.p8, p, len); |
772 | } |
773 | else |
774 | { |
775 | return equalsImpl(ptrs.p16, p, len); |
776 | } |
777 | } |
778 | |
779 | boolbool String::equals(Stringp that) const |
780 | { |
781 | if (this == that) |
782 | return truetrue; |
783 | |
784 | int32_t const len1 = this->length(); |
785 | int32_t const len2 = that->length(); |
786 | if (len1 != len2) |
787 | return falsefalse; |
788 | |
789 | Width const w1 = this->getWidth(); |
790 | Width const w2 = that->getWidth(); |
791 | |
792 | Pointers thisbuf(this); |
793 | Pointers thatbuf(that); |
794 | |
795 | switch ((w1 << 1) + w2) |
796 | { |
797 | case (k8 << 1) + k8: |
798 | return equalsImpl(thisbuf.p8, thatbuf.p8, len1); |
799 | |
800 | case (k8 << 1) + k16: |
801 | return equalsImpl(thisbuf.p8, thatbuf.p16, len1); |
802 | |
803 | case (k16 << 1) + k8: |
804 | return equalsImpl(thisbuf.p16, thatbuf.p8, len1); |
805 | |
806 | case (k16 << 1) + k16: |
807 | return equalsImpl(thisbuf.p16, thatbuf.p16, len1); |
808 | } |
809 | AvmAssert(0)do { } while (0); |
810 | return truetrue; |
811 | } |
812 | |
813 | #ifdef VMCFG_NANOJIT |
814 | /*static*/ int32_t String::equalsWithNullChecks(Stringp s1, Stringp s2) |
815 | { |
816 | if (s1 == s2) |
817 | return truetrue; |
818 | |
819 | if (!s1 || !s2) |
820 | return falsefalse; |
821 | |
822 | int32_t const len1 = s1->length(); |
823 | int32_t const len2 = s2->length(); |
824 | if (len1 != len2) |
825 | return falsefalse; |
826 | |
827 | Width const w1 = s1->getWidth(); |
828 | Width const w2 = s2->getWidth(); |
829 | |
830 | Pointers thisbuf(s1); |
831 | Pointers thatbuf(s2); |
832 | |
833 | switch ((w1 << 1) + w2) |
834 | { |
835 | case (k8 << 1) + k8: |
836 | return equalsImpl(thisbuf.p8, thatbuf.p8, len1); |
837 | |
838 | case (k8 << 1) + k16: |
839 | return equalsImpl(thisbuf.p8, thatbuf.p16, len1); |
840 | |
841 | case (k16 << 1) + k8: |
842 | return equalsImpl(thisbuf.p16, thatbuf.p8, len1); |
843 | |
844 | case (k16 << 1) + k16: |
845 | return equalsImpl(thisbuf.p16, thatbuf.p16, len1); |
846 | } |
847 | AvmAssert(0)do { } while (0); |
848 | return truetrue; |
849 | } |
850 | #endif // VMCFG_NANOJIT |
851 | |
852 | /////////////////////////////// Hash Codes ///////////////////////////////// |
853 | |
854 | // The hashing algorithm uses the full character width |
855 | |
856 | int32_t String::hashCodeLatin1(const char* buf, int32_t len) |
857 | { |
858 | return hashCodeImpl((const utf8_t*)buf, len); |
859 | } |
860 | |
861 | int32_t String::hashCodeUTF16(const wchar* buf, int32_t len) |
862 | { |
863 | return hashCodeImpl(buf, len); |
864 | } |
865 | |
866 | int32_t String::hashCodeUInt(uint32_t value) |
867 | { |
868 | // Below is largely partial-evaluated composition of |
869 | // MathUtils::convertIntegerToStringBuffer and |
870 | // String::hashCodeLatin1 |
871 | |
872 | int32_t radix = 10; |
873 | |
874 | int32_t hashCode = 0; |
875 | |
876 | int32_t len = stringLengthUInt(value); |
877 | |
878 | if (value == 0) |
879 | { |
880 | hashCode = hashCodeAddChar(hashCode, '0'); |
881 | } |
882 | else |
883 | { |
884 | uintptr_t uVal = (uintptr_t)value; |
885 | uintptr_t factor = 1; |
886 | uintptr_t cursor = len; |
887 | while (--cursor) |
888 | factor *= radix; |
889 | cursor = len; |
890 | |
891 | while (cursor > 0) |
892 | { |
893 | uintptr_t j = uVal; |
894 | j /= factor; |
895 | uVal %= factor; |
896 | factor /= radix; |
897 | AvmAssert( j < 10 )do { } while (0); |
898 | hashCode = hashCodeAddChar(hashCode, (char)(j+'0')); |
899 | cursor--; |
900 | } |
901 | } |
902 | |
903 | return hashCode; |
904 | } |
905 | |
906 | int32_t String::hashCode() const |
907 | { |
908 | if (m_length != 0) |
909 | { |
910 | Pointers ptrs(this); |
911 | if (getWidth() == k8) |
912 | { |
913 | return hashCodeImpl(ptrs.p8, m_length); |
914 | } |
915 | else |
916 | { |
917 | return hashCodeImpl(ptrs.p16, m_length); |
918 | } |
919 | } |
920 | return 0; |
921 | } |
922 | |
923 | //////////////////////////////// Accessors ///////////////////////////////// |
924 | |
925 | wchar FASTCALL__attribute__((fastcall)) String::charAt(int32_t index) const |
926 | { |
927 | AvmAssert(index >= 0 && index < m_length)do { } while (0); |
928 | |
929 | Pointers ptrs(this); |
930 | return (getWidth() == k8) ? ptrs.p8[index] : ptrs.p16[index]; |
931 | } |
932 | |
933 | int32_t String::indexOf(Stringp substr, int32_t start) const |
934 | { |
935 | if (substr == NULL__null) |
936 | return -1; |
937 | |
938 | // bug 78346: argv[1] might be less than zero. |
939 | // We clamp it to zero for two reasons: |
940 | // 1. A movie created prior to this fix with a small negative value probably worked, |
941 | // so let's fix it without breaking them. |
942 | // 2. I am told this is what java does. |
943 | if (start < 0) |
944 | start = 0; |
945 | |
946 | int32_t len = this->length(); |
947 | |
948 | if (start > len) |
949 | start = len; |
950 | |
951 | int32_t sublen = substr->length(); |
952 | |
953 | if (sublen == 0) |
954 | return start; |
955 | |
956 | // right is the last character in selfString subStr could be found at |
957 | // (and further, and there isn't enough of selfString remaining for a match to be possible) |
958 | const int32_t right = len - sublen; |
959 | if (right < 0) |
960 | return -1; |
961 | |
962 | Width const w1 = getWidth(); |
963 | Width const w2 = substr->getWidth(); |
964 | |
965 | Pointers selfBuf(this); |
966 | Pointers subBuf(substr); |
967 | |
968 | // For maximum performance, use different cases for k8/k16 combinations |
969 | if (sublen == 1) |
970 | { |
971 | switch ((w1 << 1) + w2) |
972 | { |
973 | case (k8 << 1) + k8: |
974 | return indexOfCharCodeImpl(selfBuf.p8, start, right, subBuf.p8[0]); |
975 | |
976 | case (k8 << 1) + k16: |
977 | return indexOfCharCodeImpl(selfBuf.p8, start, right, subBuf.p16[0]); |
978 | |
979 | case (k16 << 1) + k8: |
980 | return indexOfCharCodeImpl(selfBuf.p16, start, right, subBuf.p8[0]); |
981 | |
982 | case (k16 << 1) + k16: |
983 | return indexOfCharCodeImpl(selfBuf.p16, start, right, subBuf.p16[0]); |
984 | } |
985 | } |
986 | else |
987 | { |
988 | switch ((w1 << 1) + w2) |
989 | { |
990 | case (k8 << 1) + k8: |
991 | return indexOfImpl(selfBuf.p8, start, right, subBuf.p8, sublen); |
992 | |
993 | case (k8 << 1) + k16: |
994 | return indexOfImpl(selfBuf.p8, start, right, subBuf.p16, sublen); |
995 | |
996 | case (k16 << 1) + k8: |
997 | return indexOfImpl(selfBuf.p16, start, right, subBuf.p8, sublen); |
998 | |
999 | case (k16 << 1) + k16: |
1000 | return indexOfImpl(selfBuf.p16, start, right, subBuf.p16, sublen); |
1001 | } |
1002 | } |
1003 | AvmAssert(0)do { } while (0); |
1004 | return -1; |
1005 | } |
1006 | |
1007 | int32_t String::lastIndexOf(Stringp substr, int32_t start) const |
1008 | { |
1009 | if (substr == NULL__null) |
1010 | return -1; |
1011 | |
1012 | // lastIndexOf("anything", negative-number) can't match anything, ever: |
1013 | // match FP10's behavior and return -1 immediately. |
1014 | if (start < 0) |
1015 | return -1; |
1016 | |
1017 | int32_t len = this->length(); |
1018 | |
1019 | if (start > len) |
1020 | start = len; |
1021 | |
1022 | int32_t sublen = substr->length(); |
1023 | |
1024 | if (sublen == 0) |
1025 | return start; |
1026 | |
1027 | // Any match starting after right must necessarily fail, as there will be |
1028 | // insufficient characters remaining to match. |
1029 | const int32_t right = len - sublen; |
1030 | |
1031 | // The substring is longer than selfString, and therefore no match is possible. |
1032 | // We must avoid setting start to a negative value below. |
1033 | if (right < 0) |
1034 | return -1; |
1035 | |
1036 | // bug 78346: argv[1] might be greater than right |
1037 | // (similar reasons to above apply). |
1038 | if (start > right) |
1039 | start = right; |
1040 | |
1041 | Width w1 = getWidth(); |
1042 | Width w2 = substr->getWidth(); |
1043 | |
1044 | Pointers selfBuf(this); |
1045 | Pointers subBuf(substr); |
1046 | |
1047 | // For maximum performance, use different cases for k8/k16 combinations |
1048 | switch ((w1 << 1) + w2) |
1049 | { |
1050 | case (k8 << 1) + k8: |
1051 | return lastIndexOfImpl(selfBuf.p8, start, subBuf.p8, sublen); |
1052 | |
1053 | case (k8 << 1) + k16: |
1054 | return lastIndexOfImpl(selfBuf.p8, start, subBuf.p16, sublen); |
1055 | |
1056 | case (k16 << 1) + k8: |
1057 | return lastIndexOfImpl(selfBuf.p16, start, subBuf.p8, sublen); |
1058 | |
1059 | case (k16 << 1) + k16: |
1060 | return lastIndexOfImpl(selfBuf.p16, start, subBuf.p16, sublen); |
1061 | } |
1062 | AvmAssert(0)do { } while (0); |
1063 | return -1; |
1064 | } |
1065 | |
1066 | int32_t String::indexOfLatin1(const char* p, int32_t sublen, int32_t start, int32_t end) const |
1067 | { |
1068 | if (start < 0) |
1069 | start = 0; |
1070 | if (end < 0) |
1071 | end = 0; |
1072 | if (end > m_length) |
1073 | end = m_length; |
1074 | |
1075 | if (p == NULL__null || end <= start) |
1076 | return -1; |
1077 | |
1078 | if (sublen < 0) |
1079 | sublen = Length(p); |
1080 | |
1081 | if (sublen == 0) |
1082 | return start; |
1083 | |
1084 | int32_t right = end - sublen; |
1085 | if (right < 0) |
1086 | return -1; |
1087 | |
1088 | Width w = getWidth(); |
1089 | |
1090 | Pointers selfBuf(this); |
1091 | if (w == k8) |
1092 | { |
1093 | return indexOfImpl(selfBuf.p8, start, right, (const uint8_t*)p, sublen); |
1094 | } |
1095 | else |
1096 | { |
1097 | return indexOfImpl(selfBuf.p16, start, right, (const uint8_t*)p, sublen); |
1098 | } |
1099 | } |
1100 | |
1101 | int32_t String::indexOfCharCode(wchar c, int32_t start, int32_t end) const |
1102 | { |
1103 | if (start < 0) |
1104 | start = 0; |
1105 | if (end < 0) |
1106 | end = 0; |
1107 | if (end > m_length) |
1108 | end = m_length; |
1109 | |
1110 | if (end <= start) |
1111 | return -1; |
1112 | |
1113 | int32_t right = end - 1; |
1114 | if (right < 0) |
1115 | return -1; |
1116 | |
1117 | Pointers ptrs(this); |
1118 | if (getWidth() == String::k8) |
1119 | { |
1120 | return indexOfCharCodeImpl(ptrs.p8, start, right, c); |
1121 | } |
1122 | else |
1123 | { |
1124 | return indexOfCharCodeImpl(ptrs.p16, start, right, c); |
1125 | } |
1126 | } |
1127 | |
1128 | boolbool String::matchesLatin1(const char* p, int32_t len, int32_t pos) |
1129 | { |
1130 | if (p == NULL__null || pos >= m_length) |
1131 | return falsefalse; |
1132 | |
1133 | if (pos < 0) |
1134 | pos = 0; |
1135 | if (len < 0) |
1136 | len = Length(p); |
1137 | |
1138 | Pointers ptrs(this); |
1139 | if (getWidth() == k8) |
1140 | { |
1141 | return equalsImpl(ptrs.p8 + pos, (const uint8_t*)p, len); |
1142 | } |
1143 | else |
1144 | { |
1145 | return equalsImpl(ptrs.p16 + pos, (const uint8_t*)p, len); |
1146 | } |
1147 | } |
1148 | |
1149 | boolbool String::matchesLatin1_caseless(const char* p, int32_t len, int32_t pos) |
1150 | { |
1151 | if (p == NULL__null || pos >= m_length) |
1152 | return falsefalse; |
1153 | |
1154 | if (pos < 0) |
1155 | pos = 0; |
1156 | if (len < 0) |
1157 | len = Length(p); |
1158 | |
1159 | StringIndexer self(this); |
1160 | while (len--) |
1161 | { |
1162 | wchar const ch1 = wCharToUpper(self[pos++]); |
1163 | wchar const ch2 = wCharToUpper(uint8_t(*p++)); |
1164 | if (ch1 != ch2) |
1165 | return falsefalse; |
1166 | } |
1167 | return truetrue; |
1168 | } |
1169 | |
1170 | ///////////////////////////// Concatenation //////////////////////////////// |
1171 | |
1172 | /*static*/ Stringp String::concatStrings(Stringp leftStr, Stringp rightStr) |
1173 | { |
1174 | if (leftStr == NULL__null || leftStr->m_length == 0) |
1175 | return rightStr; |
1176 | |
1177 | return leftStr->_append(rightStr, Pointers(rightStr), rightStr->length(), rightStr->getWidth()); |
1178 | } |
1179 | |
1180 | Stringp String::append(Stringp rightStr) |
1181 | { |
1182 | if (rightStr == NULL__null || rightStr->m_length == 0) |
1183 | return this; |
1184 | |
1185 | return _append(rightStr, Pointers(rightStr), rightStr->length(), rightStr->getWidth()); |
1186 | } |
1187 | |
1188 | // NB: if rightStrPtr is nonnull, it is assumed to be the source of rightStr. |
1189 | // In this case, rightStr is assumed to be invalidate by any possible GC activity, |
1190 | // and will be re-created from rightStrPtr. |
1191 | Stringp String::_append(Stringp rightStrPtr, const Pointers& rightStr, int32_t numChars, Width charWidth) |
1192 | { |
1193 | if (numChars <= 0) |
1194 | return this; |
1195 | |
1196 | AvmAssert(charWidth != kAuto)do { } while (0); |
1197 | |
1198 | MMgc::GC* gc = _gc(this); |
1199 | // fromCharCode() optimization: If this string is empty, and the buffer is a single |
1200 | // ASCII character, return the cached character |
1201 | if (m_length == 0 && numChars == 1) |
1202 | { |
1203 | // Sun studio generated wrong code for the following Conditional operator. |
1204 | // Add type conversion to wchar as a workaround. |
1205 | wchar ch = (charWidth == k8) ? (wchar)rightStr.p8[0] : (wchar)rightStr.p16[0]; |
1206 | if (ch < 128) |
1207 | return gc->core()->cachedChars[ch]; |
1208 | } |
1209 | |
1210 | Width thisWidth = getWidth(); |
1211 | Width newWidth = (thisWidth < charWidth) ? charWidth : thisWidth; |
1212 | |
1213 | int32_t newLen = int32AddChecked(m_length, numChars); |
1214 | |
1215 | Stringp master = (isDependent()) ? m_extra.master : this; |
1216 | // check for characters left in leftStr's buffer, or if leftStr has spent its padding already |
1217 | // string types other than kDynamic have charsLeft == 0 |
1218 | int32_t charsLeft = 0; |
1219 | int32_t charsUsed = 0; |
1220 | if (thisWidth >= charWidth) |
1221 | { |
1222 | // in-place append only if rightStr's width fits into leftStr |
1223 | charsLeft = master->getCharsLeft(); |
1224 | if (!master->isStatic()) |
1225 | // charsUsed is the number of chars already spent behind m_length |
1226 | // Usage of m_buffer is OK - master can never be dependent |
1227 | charsUsed = int32_t((GC::Size(master->m_buffer.pv) >> thisWidth) - master->m_length - charsLeft); |
1228 | } |
1229 | int32_t start = 0; // string start for dependent strings |
1230 | |
1231 | // it is possible to append in-place if |
1232 | // 1) this is a kDynamic string and charsUsed == 0 |
1233 | // 2) this is a kDependent string and the end matches the master's real end |
1234 | switch (getType()) |
1235 | { |
1236 | case kDynamic: |
1237 | if (charsUsed != 0) |
1238 | // someone else has already appended in-place |
1239 | charsLeft = 0; |
1240 | break; |
1241 | case kDependent: |
1242 | start = (int32_t) m_buffer.offset_bytes >> thisWidth; |
1243 | if ((start + m_length) != master->m_length + charsUsed) |
1244 | charsLeft = 0; |
1245 | break; |
1246 | default:; // kStatic |
1247 | } |
1248 | |
1249 | // the big check: are there enough chars left? |
1250 | if (numChars <= charsLeft) |
1251 | { |
1252 | // the right-hand string fits into the buffer end |
1253 | _copyBuffers(rightStr.pv, |
1254 | Pointers(this).p8 + (m_length << thisWidth), // m_length << thiswidth is safe by definition |
1255 | numChars, charWidth, newWidth); |
1256 | |
1257 | charsUsed += numChars; |
Value stored to 'charsUsed' is never read | |
1258 | charsLeft -= numChars; |
1259 | master->setCharsLeft(charsLeft); |
1260 | #ifdef _DEBUG |
1261 | // Terminate string with 0 for better debugging display |
1262 | if (charsLeft && !master->isStatic()) |
1263 | { |
1264 | int32_t end = master->m_length + charsUsed; |
1265 | if (newWidth == k8) |
1266 | master->m_buffer.p8[end] = 0; |
1267 | else |
1268 | master->m_buffer.p16[end] = 0; |
1269 | } |
1270 | #endif |
1271 | Stringp s = createDependent(gc, master, start, newLen); |
1272 | // createDependent propagates the 7-bit flag, which is a good idea for substrings. |
1273 | // however, if we are adding data, it may no longer qualify. rather than checking |
1274 | // the appended data, just clear the flag (it will be lazily recalculated as needed) |
1275 | s->m_bitsAndFlags &= ~TSTR_7BIT_FLAG; |
1276 | return s; |
1277 | } |
1278 | |
1279 | // fall thru - string does not fit |
1280 | // create a new kDynamic string containing the concatenated string |
1281 | // See the definition of TSTR_MAX_CHARSLEFT above for an explanation |
1282 | // of this algorithm |
1283 | int32_t newSize = (newLen < TSTR_MIN_DYNAMIC_ALLOCATION32) ? TSTR_MIN_DYNAMIC_ALLOCATION32 : int32ShlChecked(newLen, 1); |
1284 | int32_t extra = newSize - newLen; |
1285 | if (extra > (int32_t) TSTR_MAX_CHARSLEFT((uint32_t)TSTR_CHARSLEFT_MASK >> TSTR_CHARSLEFT_SHIFT >> 1)) |
1286 | extra = (int32_t) TSTR_MAX_CHARSLEFT((uint32_t)TSTR_CHARSLEFT_MASK >> TSTR_CHARSLEFT_SHIFT >> 1); |
1287 | |
1288 | const boolbool is7bit = falsefalse; // actually, might be, we just haven't checked yet. |
1289 | Stringp newStr = createDynamic(gc, NULL__null, newLen, newWidth, is7bit, extra); |
1290 | |
1291 | // note that createDynamic has invalidated any existing Pointers structs... |
1292 | const void* srcLeft = Pointers(this).pv; |
1293 | const void* srcRight = rightStrPtr ? Pointers(rightStrPtr).pv : rightStr.pv; |
1294 | |
1295 | // copy leftStr |
1296 | void* ptr = _copyBuffers(srcLeft, |
1297 | newStr->m_buffer.pv, |
1298 | m_length, |
1299 | thisWidth, newWidth); |
1300 | |
1301 | // append src |
1302 | _copyBuffers(srcRight, |
1303 | ptr, |
1304 | numChars, |
1305 | charWidth, newWidth); |
1306 | |
1307 | #ifdef _DEBUG |
1308 | // Terminate string with 0 for better debugging display |
1309 | if (newStr->getCharsLeft()) |
1310 | { |
1311 | if (newWidth == k8) |
1312 | newStr->m_buffer.p8[newStr->m_length] = 0; |
1313 | else |
1314 | newStr->m_buffer.p16[newStr->m_length] = 0; |
1315 | } |
1316 | #endif |
1317 | VERIFY_7BIT(newStr)do { } while (0); |
1318 | return newStr; |
1319 | } |
1320 | |
1321 | /////////////////////////////// substrings ///////////////////////////////// |
1322 | |
1323 | // Remember: string runs from start up to, but not including, the end! |
1324 | Stringp String::substring(int32_t start, int32_t end) |
1325 | { |
1326 | if (start < 0) |
1327 | start = 0; |
1328 | else if (start > m_length) |
1329 | start = m_length; |
1330 | if (end < 0) |
1331 | end = 0; |
1332 | else if (end > m_length) |
1333 | end = m_length; |
1334 | |
1335 | // does the substring span this string? |
1336 | if (start == 0 && end == m_length) |
1337 | return this; |
1338 | |
1339 | MMgc::GC* gc = _gc(this); |
1340 | AvmCore* core = gc->core(); |
1341 | // for empty strings, return the standard Empty String instance |
1342 | if (end <= start) |
1343 | return core->kEmptyString; |
1344 | |
1345 | // for single characters < 128, return the cached string |
1346 | if (end == (start + 1)) |
1347 | { |
1348 | wchar const ch = charAt(start); |
1349 | if (ch < 128) |
1350 | return core->cachedChars[ch]; |
1351 | } |
1352 | |
1353 | // otherwise, create a dependent string |
1354 | Stringp master = this; |
1355 | if (isDependent()) |
1356 | { |
1357 | // get the string offset |
1358 | master = m_extra.master; |
1359 | // TODO: possible 32-bit overflow for a very huge dependent string |
1360 | int32_t offset = (int32_t) m_buffer.offset_bytes >> getWidth(); |
1361 | start += (int32_t) offset; |
1362 | end += (int32_t) offset; |
1363 | } |
1364 | AvmAssert(!master->isDependent())do { } while (0); |
1365 | // you may be tempted to optimize for static strings, by creating |
1366 | // a new static string pointing to the substring... do not do this. |
1367 | // it will break static strings that point directly into ABC buffers. |
1368 | return createDependent(gc, master, start, end - start); |
1369 | } |
1370 | |
1371 | Stringp String::substr(int32_t start, int32_t len) |
1372 | { |
1373 | start = (int32_t) NativeObjectHelpers::ClampIndexInt(start, m_length); |
1374 | len = (int32_t) NativeObjectHelpers::ClampIndexInt(len, m_length); |
1375 | int32_t end = (m_length-len<=start)? m_length:start+len; // no need to further clamp |
1376 | return substring(start, end); |
1377 | } |
1378 | |
1379 | Stringp String::slice(int32_t start, int32_t end) |
1380 | { |
1381 | start = (int32_t) NativeObjectHelpers::ClampIndexInt(start, m_length); |
1382 | end = (int32_t) NativeObjectHelpers::ClampIndexInt(end, m_length); |
1383 | if (end < start) |
1384 | end = start; |
1385 | |
1386 | return substr(start, end-start); |
1387 | } |
1388 | |
1389 | ////////////////////////////////// Parsing //////////////////////////////////// |
1390 | |
1391 | // This routine is a very specific parser to generate a positive integer from a string. |
1392 | // The following are supported: |
1393 | // "0" - one single digit for zero - NOT "00" or any other form of zero |
1394 | //[1-9]+[0-9]* up to 2^32-2 (4294967294) |
1395 | // 2^32-1 (4294967295) is not supported (see ECMA quote below). |
1396 | // The ECMA that we're supporting with this routine is... |
1397 | // cn: the ES3 test for a valid array index is |
1398 | // "A property name P (in the form of a string value) is an array index if and |
1399 | // only if ToString(ToUint32(P)) is equal to P and ToUint32(P) is not equal to 2^32-1." |
1400 | |
1401 | // Don't support 000000 as 0. |
1402 | // We don't support 0x1234 as 1234 in hex since string(1234) doesn't equal '0x1234') |
1403 | // No leading zeros are supported |
1404 | |
1405 | boolbool String::parseIndex(uint32_t& result) const |
1406 | { |
1407 | // avoid parsing if this string is definitely not a candidate |
1408 | // A string that is more than 10 digits (and does not start with 0) |
1409 | // will always be greater than 2^32-1 (4294967295) or not a numeric string |
1410 | if (m_bitsAndFlags & TSTR_NOUINT_FLAG) |
1411 | return falsefalse; |
1412 | |
1413 | if (m_bitsAndFlags & (TSTR_UINT28_FLAG | TSTR_UINT32_FLAG)) |
1414 | { |
1415 | result = m_extra.index; |
1416 | return truetrue; |
1417 | } |
1418 | |
1419 | int64_t n = 0; |
1420 | wchar wch; |
1421 | Pointers ptrs(this); |
1422 | Width w = getWidth(); |
1423 | if (m_length == 0 || m_length > 10) |
1424 | goto bad; |
1425 | |
1426 | // collect the value |
1427 | for (int32_t i = 0; i < m_length; i++) |
1428 | { |
1429 | if (w == k8) |
1430 | wch = wchar(*ptrs.p8++); |
1431 | else |
1432 | wch = *ptrs.p16++; |
1433 | // bad character, or leading zero? |
1434 | if (wch < '0' || wch > '9' || (i == 1 && n == 0)) |
1435 | goto bad; |
1436 | n = (n * 10) + wch - '0'; |
1437 | } |
1438 | // out of range? |
1439 | if (n >= 0xFFFFFFFF) |
1440 | goto bad; |
1441 | if (n & ScriptObject::MAX_INTEGER_MASK) |
1442 | m_bitsAndFlags |= TSTR_NOINT_FLAG; |
1443 | if (!isDependent()) |
1444 | { |
1445 | m_bitsAndFlags |= TSTR_UINT32_FLAG; |
1446 | m_extra.index = uint32_t(n); |
1447 | } |
1448 | result = uint32_t(n); |
1449 | return truetrue; |
1450 | bad: |
1451 | m_bitsAndFlags |= TSTR_NOINT_FLAG | TSTR_NOUINT_FLAG; |
1452 | return falsefalse; |
1453 | } |
1454 | |
1455 | Atom String::getIntAtom() const |
1456 | { |
1457 | // The call uses a simplified version of parseIndex with 32-bit arithmetic, |
1458 | // since we can only fit 28 bits. |
1459 | if (m_bitsAndFlags & TSTR_NOINT_FLAG) |
1460 | return 0; |
1461 | |
1462 | if (m_bitsAndFlags & TSTR_UINT28_FLAG) |
1463 | return Atom((m_extra.index << 3) | AtomConstants::kIntptrType); |
1464 | |
1465 | int32_t n = 0; |
1466 | wchar wch; |
1467 | Pointers ptrs(this); |
1468 | Width w = getWidth(); |
1469 | if (m_length == 0 || m_length > 10) |
1470 | goto bad; |
1471 | |
1472 | // collect the value |
1473 | for (int32_t i = 0; i < m_length; i++) |
1474 | { |
1475 | if (w == k8) |
1476 | wch = wchar(*ptrs.p8++); |
1477 | else |
1478 | wch = *ptrs.p16++; |
1479 | // bad character, or leading zero? |
1480 | if (wch < '0' || wch > '9' || (i == 1 && n == 0)) |
1481 | goto bad; |
1482 | n = (n * 10) + wch - '0'; |
1483 | if (n & ScriptObject::MAX_INTEGER_MASK) |
1484 | goto bad; |
1485 | } |
1486 | if (!isDependent()) |
1487 | { |
1488 | m_bitsAndFlags |= TSTR_UINT28_FLAG; |
1489 | m_extra.index = n; |
1490 | } |
1491 | return Atom((n << 3) | AtomConstants::kIntptrType); |
1492 | bad: |
1493 | m_bitsAndFlags |= TSTR_NOINT_FLAG; |
1494 | return 0; |
1495 | } |
1496 | |
1497 | /////////////////////////////// Case conversion ///////////////////////////////// |
1498 | |
1499 | |
1500 | static const wchar lowerCaseBase[] = |
1501 | { |
1502 | /*0x0061, 0x0062, 0x0063, 0x0064, 0x0065, 0x0066, 0x0067, 0x0068, 0x0069, 0x006A, |
1503 | 0x006B, 0x006C, 0x006D, 0x006E, 0x006F, 0x0070, 0x0071, 0x0072, 0x0073, 0x0074, |
1504 | 0x0075, 0x0076, 0x0077, 0x0078, 0x0079, 0x007A,*/ 0x00B5, /*0x00E0, 0x00E1, 0x00E2, |
1505 | 0x00E3, 0x00E4, 0x00E5, 0x00E6, 0x00E7, 0x00E8, 0x00E9, 0x00EA, 0x00EB, 0x00EC, |
1506 | 0x00ED, 0x00EE, 0x00EF, 0x00F0, 0x00F1, 0x00F2, 0x00F3, 0x00F4, 0x00F5, 0x00F6, |
1507 | 0x00F8, 0x00F9, 0x00FA, 0x00FB, 0x00FC, 0x00FD, 0x00FE,*/ 0x00FF, /*0x0101, 0x0103, |
1508 | 0x0105, 0x0107, 0x0109, 0x010B, 0x010D, 0x010F, 0x0111, 0x0113, 0x0115, 0x0117, |
1509 | 0x0119, 0x011B, 0x011D, 0x011F, 0x0121, 0x0123, 0x0125, 0x0127, 0x0129, 0x012B, |
1510 | 0x012D, 0x012F,*/ 0x0131, 0x0133, 0x0135, 0x0137, /*0x013A, 0x013C, 0x013E, 0x0140, |
1511 | 0x0142, 0x0144, 0x0146, 0x0148,*/ /*0x014B, 0x014D, 0x014F, 0x0151, 0x0153, 0x0155, |
1512 | 0x0157, 0x0159, 0x015B, 0x015D, 0x015F, 0x0161, 0x0163, 0x0165, 0x0167, 0x0169, |
1513 | 0x016B, 0x016D, 0x016F, 0x0171, 0x0173, 0x0175, 0x0177,*/ 0x017A, 0x017C, 0x017E, |
1514 | 0x017F, 0x0183, 0x0185, 0x0188, 0x018C, 0x0192, 0x0195, 0x0199, 0x01A1, 0x01A3, |
1515 | 0x01A5, 0x01A8, 0x01AD, 0x01B0, 0x01B4, 0x01B6, 0x01B9, 0x01BD, 0x01BF, 0x01C5, |
1516 | 0x01C6, 0x01C8, 0x01C9, 0x01CB, 0x01CC, 0x01CE, 0x01D0, 0x01D2, 0x01D4, 0x01D6, |
1517 | 0x01D8, 0x01DA, 0x01DC, 0x01DD, 0x01DF, 0x01E1, 0x01E3, 0x01E5, 0x01E7, 0x01E9, |
1518 | 0x01EB, 0x01ED, 0x01EF, 0x01F2, 0x01F3, 0x01F5, 0x01F9, 0x01FB, 0x01FD, 0x01FF, |
1519 | /*0x0201, 0x0203, 0x0205, 0x0207, 0x0209, 0x020B, 0x020D, 0x020F, 0x0211, 0x0213, |
1520 | 0x0215, 0x0217, 0x0219, 0x021B, 0x021D, 0x021F, 0x0223, 0x0225, 0x0227, 0x0229, |
1521 | 0x022B, 0x022D, 0x022F, 0x0231, 0x0233,*/ 0x0253, 0x0254, 0x0256, 0x0257, 0x0259, |
1522 | 0x025B, 0x0260, 0x0263, 0x0268, 0x0269, 0x026F, 0x0272, 0x0275, 0x0280, 0x0283, |
1523 | 0x0288, 0x028A, 0x028B, 0x0292, 0x0345, 0x03AC, 0x03AD, 0x03AE, 0x03AF, /*0x03B1, |
1524 | 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7, 0x03B8, 0x03B9, 0x03BA, 0x03BB, |
1525 | 0x03BC, 0x03BD, 0x03BE, 0x03BF, 0x03C0, 0x03C1, 0x03C2, 0x03C3, 0x03C4, 0x03C5, |
1526 | 0x03C6, 0x03C7, 0x03C8, 0x03C9, 0x03CA, 0x03CB,*/ 0x03CC, 0x03CD, 0x03CE, 0x03D0, |
1527 | 0x03D1, 0x03D5, 0x03D6, /*0x03DB, 0x03DD, 0x03DF, 0x03E1, 0x03E3, 0x03E5, 0x03E7, |
1528 | 0x03E9, 0x03EB, 0x03ED, 0x03EF,*/ 0x03F0, 0x03F1, 0x03F2, 0x03F5, /*0x0430, 0x0431, |
1529 | 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, 0x0437, 0x0438, 0x0439, 0x043A, 0x043B, |
1530 | 0x043C, 0x043D, 0x043E, 0x043F, 0x0440, 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, |
1531 | 0x0446, 0x0447, 0x0448, 0x0449, 0x044A, 0x044B, 0x044C, 0x044D, 0x044E, 0x044F,*/ |
1532 | /*0x0450, 0x0451, 0x0452, 0x0453, 0x0454, 0x0455, 0x0456, 0x0457, 0x0458, 0x0459, |
1533 | 0x045A, 0x045B, 0x045C, 0x045D, 0x045E, 0x045F,*/ /*0x0461, 0x0463, 0x0465, 0x0467, |
1534 | 0x0469, 0x046B, 0x046D, 0x046F, 0x0471, 0x0473, 0x0475, 0x0477, 0x0479, 0x047B, |
1535 | 0x047D, 0x047F, 0x0481, 0x048D, 0x048F, 0x0491, 0x0493, 0x0495, 0x0497, 0x0499, |
1536 | 0x049B, 0x049D, 0x049F, 0x04A1, 0x04A3, 0x04A5, 0x04A7, 0x04A9, 0x04AB, 0x04AD, |
1537 | 0x04AF, 0x04B1, 0x04B3, 0x04B5, 0x04B7, 0x04B9, 0x04BB, 0x04BD, 0x04BF,*/ 0x04C2, |
1538 | 0x04C4, 0x04C8, 0x04CC, /*0x04D1, 0x04D3, 0x04D5, 0x04D7, 0x04D9, 0x04DB, 0x04DD, |
1539 | 0x04DF, 0x04E1, 0x04E3, 0x04E5, 0x04E7, 0x04E9, 0x04EB, 0x04ED, 0x04EF, 0x04F1, |
1540 | 0x04F3, 0x04F5, 0x04F9, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567, |
1541 | 0x0568, 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F, 0x0570, 0x0571, |
1542 | 0x0572, 0x0573, 0x0574, 0x0575, 0x0576, 0x0577, 0x0578, 0x0579, 0x057A, 0x057B, |
1543 | 0x057C, 0x057D, 0x057E, 0x057F, 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, |
1544 | 0x0586, 0x1E01, 0x1E03, 0x1E05, 0x1E07, 0x1E09, 0x1E0B, 0x1E0D, 0x1E0F, 0x1E11, |
1545 | 0x1E13, 0x1E15, 0x1E17, 0x1E19, 0x1E1B, 0x1E1D, 0x1E1F, 0x1E21, 0x1E23, 0x1E25, |
1546 | 0x1E27, 0x1E29, 0x1E2B, 0x1E2D, 0x1E2F, 0x1E31, 0x1E33, 0x1E35, 0x1E37, 0x1E39, |
1547 | 0x1E3B, 0x1E3D, 0x1E3F, 0x1E41, 0x1E43, 0x1E45, 0x1E47, 0x1E49, 0x1E4B, 0x1E4D, |
1548 | 0x1E4F, 0x1E51, 0x1E53, 0x1E55, 0x1E57, 0x1E59, 0x1E5B, 0x1E5D, 0x1E5F, 0x1E61, |
1549 | 0x1E63, 0x1E65, 0x1E67, 0x1E69, 0x1E6B, 0x1E6D, 0x1E6F, 0x1E71, 0x1E73, 0x1E75, |
1550 | 0x1E77, 0x1E79, 0x1E7B, 0x1E7D, 0x1E7F, 0x1E81, 0x1E83, 0x1E85, 0x1E87, 0x1E89, |
1551 | 0x1E8B, 0x1E8D, 0x1E8F, 0x1E91, 0x1E93, 0x1E95,*/ 0x1E9B, /*0x1EA1, 0x1EA3, 0x1EA5, |
1552 | 0x1EA7, 0x1EA9, 0x1EAB, 0x1EAD, 0x1EAF, 0x1EB1, 0x1EB3, 0x1EB5, 0x1EB7, 0x1EB9, |
1553 | 0x1EBB, 0x1EBD, 0x1EBF, 0x1EC1, 0x1EC3, 0x1EC5, 0x1EC7, 0x1EC9, 0x1ECB, 0x1ECD, |
1554 | 0x1ECF, 0x1ED1, 0x1ED3, 0x1ED5, 0x1ED7, 0x1ED9, 0x1EDB, 0x1EDD, 0x1EDF, 0x1EE1, |
1555 | 0x1EE3, 0x1EE5, 0x1EE7, 0x1EE9, 0x1EEB, 0x1EED, 0x1EEF, 0x1EF1, 0x1EF3, 0x1EF5, |
1556 | 0x1EF7, 0x1EF9,*/ /*0x1F00, 0x1F01, 0x1F02, 0x1F03, 0x1F04, 0x1F05, 0x1F06, 0x1F07,*/ |
1557 | /*0x1F10, 0x1F11, 0x1F12, 0x1F13, 0x1F14, 0x1F15, 0x1F20, 0x1F21, 0x1F22, 0x1F23, |
1558 | 0x1F24, 0x1F25, 0x1F26, 0x1F27, 0x1F30, 0x1F31, 0x1F32, 0x1F33, 0x1F34, 0x1F35, |
1559 | 0x1F36, 0x1F37,*/ 0x1F40, 0x1F41, 0x1F42, 0x1F43, 0x1F44, 0x1F45, 0x1F51, 0x1F53, |
1560 | 0x1F55, 0x1F57, 0x1F60, 0x1F61, 0x1F62, 0x1F63, 0x1F64, 0x1F65, 0x1F66, 0x1F67, |
1561 | 0x1F70, 0x1F71, 0x1F72, 0x1F73, 0x1F74, 0x1F75, 0x1F76, 0x1F77, 0x1F78, 0x1F79, |
1562 | 0x1F7A, 0x1F7B, 0x1F7C, 0x1F7D, 0x1F80, 0x1F81, 0x1F82, 0x1F83, 0x1F84, 0x1F85, |
1563 | 0x1F86, 0x1F87, 0x1F90, 0x1F91, 0x1F92, 0x1F93, 0x1F94, 0x1F95, 0x1F96, 0x1F97, |
1564 | 0x1FA0, 0x1FA1, 0x1FA2, 0x1FA3, 0x1FA4, 0x1FA5, 0x1FA6, 0x1FA7, 0x1FB0, 0x1FB1, |
1565 | 0x1FB3, 0x1FBE, 0x1FC3, 0x1FD0, 0x1FD1, 0x1FE0, 0x1FE1, 0x1FE5, 0x1FF3 /*0x2170, |
1566 | 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177, 0x2178, 0x2179, 0x217A, |
1567 | 0x217B, 0x217C, 0x217D, 0x217E, 0x217F, 0x24D0, 0x24D1, 0x24D2, 0x24D3, 0x24D4, |
1568 | 0x24D5, 0x24D6, 0x24D7, 0x24D8, 0x24D9, 0x24DA, 0x24DB, 0x24DC, 0x24DD, 0x24DE, |
1569 | 0x24DF, 0x24E0, 0x24E1, 0x24E2, 0x24E3, 0x24E4, 0x24E5, 0x24E6, 0x24E7, 0x24E8, |
1570 | 0x24E9 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47, 0xFF48, 0xFF49, |
1571 | 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F, 0xFF50, 0xFF51, 0xFF52, 0xFF53, |
1572 | 0xFF54, 0xFF55, 0xFF56, 0xFF57, 0xFF58, 0xFF59, 0xFF5A*/ |
1573 | }; |
1574 | |
1575 | static const wchar upperCaseConversion[] = |
1576 | { |
1577 | /*0x0041, 0x0042, 0x0043, 0x0044, 0x0045, 0x0046, 0x0047, 0x0048, 0x0049, 0x004A, |
1578 | 0x004B, 0x004C, 0x004D, 0x004E, 0x004F, 0x0050, 0x0051, 0x0052, 0x0053, 0x0054, |
1579 | 0x0055, 0x0056, 0x0057, 0x0058, 0x0059, 0x005A,*/ 0x039C, /*0x00C0, 0x00C1, 0x00C2, |
1580 | 0x00C3, 0x00C4, 0x00C5, 0x00C6, 0x00C7, 0x00C8, 0x00C9, 0x00CA, 0x00CB, 0x00CC, |
1581 | 0x00CD, 0x00CE, 0x00CF, 0x00D0, 0x00D1, 0x00D2, 0x00D3, 0x00D4, 0x00D5, 0x00D6, |
1582 | 0x00D8, 0x00D9, 0x00DA, 0x00DB, 0x00DC, 0x00DD, 0x00DE,*/ 0x0178, /*0x0100, 0x0102, |
1583 | 0x0104, 0x0106, 0x0108, 0x010A, 0x010C, 0x010E, 0x0110, 0x0112, 0x0114, 0x0116, |
1584 | 0x0118, 0x011A, 0x011C, 0x011E, 0x0120, 0x0122, 0x0124, 0x0126, 0x0128, 0x012A, |
1585 | 0x012C, 0x012E,*/ 0x0049, 0x0132, 0x0134, 0x0136, /*0x0139, 0x013B, 0x013D, 0x013F, |
1586 | 0x0141, 0x0143, 0x0145, 0x0147,*/ /*0x014A, 0x014C, 0x014E, 0x0150, 0x0152, 0x0154, |
1587 | 0x0156, 0x0158, 0x015A, 0x015C, 0x015E, 0x0160, 0x0162, 0x0164, 0x0166, 0x0168, |
1588 | 0x016A, 0x016C, 0x016E, 0x0170, 0x0172, 0x0174, 0x0176,*/ 0x0179, 0x017B, 0x017D, |
1589 | 0x0053, 0x0182, 0x0184, 0x0187, 0x018B, 0x0191, 0x01F6, 0x0198, 0x01A0, 0x01A2, |
1590 | 0x01A4, 0x01A7, 0x01AC, 0x01AF, 0x01B3, 0x01B5, 0x01B8, 0x01BC, 0x01F7, 0x01C4, |
1591 | 0x01C4, 0x01C7, 0x01C7, 0x01CA, 0x01CA, 0x01CD, 0x01CF, 0x01D1, 0x01D3, 0x01D5, |
1592 | 0x01D7, 0x01D9, 0x01DB, 0x018E, 0x01DE, 0x01E0, 0x01E2, 0x01E4, 0x01E6, 0x01E8, |
1593 | 0x01EA, 0x01EC, 0x01EE, 0x01F1, 0x01F1, 0x01F4, 0x01F8, 0x01FA, 0x01FC, 0x01FE, |
1594 | /*0x0200, 0x0202, 0x0204, 0x0206, 0x0208, 0x020A, 0x020C, 0x020E, 0x0210, 0x0212, |
1595 | 0x0214, 0x0216, 0x0218, 0x021A, 0x021C, 0x021E, 0x0222, 0x0224, 0x0226, 0x0228, |
1596 | 0x022A, 0x022C, 0x022E, 0x0230, 0x0232,*/ 0x0181, 0x0186, 0x0189, 0x018A, 0x018F, |
1597 | 0x0190, 0x0193, 0x0194, 0x0197, 0x0196, 0x019C, 0x019D, 0x019F, 0x01A6, 0x01A9, |
1598 | 0x01AE, 0x01B1, 0x01B2, 0x01B7, 0x0399, 0x0386, 0x0388, 0x0389, 0x038A, /*0x0391, |
1599 | 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397, 0x0398, 0x0399, 0x039A, 0x039B, |
1600 | 0x039C, 0x039D, 0x039E, 0x039F, 0x03A0, 0x03A1, 0x03A3, 0x03A3, 0x03A4, 0x03A5, |
1601 | 0x03A6, 0x03A7, 0x03A8, 0x03A9, 0x03AA, 0x03AB,*/ 0x038C, 0x038E, 0x038F, 0x0392, |
1602 | 0x0398, 0x03A6, 0x03A0, /*0x03DA, 0x03DC, 0x03DE, 0x03E0, 0x03E2, 0x03E4, 0x03E6, |
1603 | 0x03E8, 0x03EA, 0x03EC, 0x03EE,*/ 0x039A, 0x03A1, 0x03A3, 0x0395, /*0x0410, 0x0411, |
1604 | 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, 0x0417, 0x0418, 0x0419, 0x041A, 0x041B, |
1605 | 0x041C, 0x041D, 0x041E, 0x041F, 0x0420, 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, |
1606 | 0x0426, 0x0427, 0x0428, 0x0429, 0x042A, 0x042B, 0x042C, 0x042D, 0x042E, 0x042F,*/ |
1607 | /*0x0400, 0x0401, 0x0402, 0x0403, 0x0404, 0x0405, 0x0406, 0x0407, 0x0408, 0x0409, |
1608 | 0x040A, 0x040B, 0x040C, 0x040D, 0x040E, 0x040F,*//*0x0460, 0x0462, 0x0464, 0x0466, |
1609 | 0x0468, 0x046A, 0x046C, 0x046E, 0x0470, 0x0472, 0x0474, 0x0476, 0x0478, 0x047A, |
1610 | 0x047C, 0x047E, 0x0480, 0x048C, 0x048E, 0x0490, 0x0492, 0x0494, 0x0496, 0x0498, |
1611 | 0x049A, 0x049C, 0x049E, 0x04A0, 0x04A2, 0x04A4, 0x04A6, 0x04A8, 0x04AA, 0x04AC, |
1612 | 0x04AE, 0x04B0, 0x04B2, 0x04B4, 0x04B6, 0x04B8, 0x04BA, 0x04BC, 0x04BE,*/ 0x04C1, |
1613 | 0x04C3, 0x04C7, 0x04CB, /*0x04D0, 0x04D2, 0x04D4, 0x04D6, 0x04D8, 0x04DA, 0x04DC, |
1614 | 0x04DE, 0x04E0, 0x04E2, 0x04E4, 0x04E6, 0x04E8, 0x04EA, 0x04EC, 0x04EE, 0x04F0, |
1615 | 0x04F2, 0x04F4, 0x04F8, 0x0531, 0x0532, 0x0533, 0x0534, 0x0535, 0x0536, 0x0537, |
1616 | 0x0538, 0x0539, 0x053A, 0x053B, 0x053C, 0x053D, 0x053E, 0x053F, 0x0540, 0x0541, |
1617 | 0x0542, 0x0543, 0x0544, 0x0545, 0x0546, 0x0547, 0x0548, 0x0549, 0x054A, 0x054B, |
1618 | 0x054C, 0x054D, 0x054E, 0x054F, 0x0550, 0x0551, 0x0552, 0x0553, 0x0554, 0x0555, |
1619 | 0x0556, 0x1E00, 0x1E02, 0x1E04, 0x1E06, 0x1E08, 0x1E0A, 0x1E0C, 0x1E0E, 0x1E10, |
1620 | 0x1E12, 0x1E14, 0x1E16, 0x1E18, 0x1E1A, 0x1E1C, 0x1E1E, 0x1E20, 0x1E22, 0x1E24, |
1621 | 0x1E26, 0x1E28, 0x1E2A, 0x1E2C, 0x1E2E, 0x1E30, 0x1E32, 0x1E34, 0x1E36, 0x1E38, |
1622 | 0x1E3A, 0x1E3C, 0x1E3E, 0x1E40, 0x1E42, 0x1E44, 0x1E46, 0x1E48, 0x1E4A, 0x1E4C, |
1623 | 0x1E4E, 0x1E50, 0x1E52, 0x1E54, 0x1E56, 0x1E58, 0x1E5A, 0x1E5C, 0x1E5E, 0x1E60, |
1624 | 0x1E62, 0x1E64, 0x1E66, 0x1E68, 0x1E6A, 0x1E6C, 0x1E6E, 0x1E70, 0x1E72, 0x1E74, |
1625 | 0x1E76, 0x1E78, 0x1E7A, 0x1E7C, 0x1E7E, 0x1E80, 0x1E82, 0x1E84, 0x1E86, 0x1E88, |
1626 | 0x1E8A, 0x1E8C, 0x1E8E, 0x1E90, 0x1E92, 0x1E94,*/ 0x1E60, /*0x1EA0, 0x1EA2, 0x1EA4, |
1627 | 0x1EA6, 0x1EA8, 0x1EAA, 0x1EAC, 0x1EAE, 0x1EB0, 0x1EB2, 0x1EB4, 0x1EB6, 0x1EB8, |
1628 | 0x1EBA, 0x1EBC, 0x1EBE, 0x1EC0, 0x1EC2, 0x1EC4, 0x1EC6, 0x1EC8, 0x1ECA, 0x1ECC, |
1629 | 0x1ECE, 0x1ED0, 0x1ED2, 0x1ED4, 0x1ED6, 0x1ED8, 0x1EDA, 0x1EDC, 0x1EDE, 0x1EE0, |
1630 | 0x1EE2, 0x1EE4, 0x1EE6, 0x1EE8, 0x1EEA, 0x1EEC, 0x1EEE, 0x1EF0, 0x1EF2, 0x1EF4, |
1631 | 0x1EF6, 0x1EF8,*/ /*0x1F08, 0x1F09, 0x1F0A, 0x1F0B, 0x1F0C, 0x1F0D, 0x1F0E, 0x1F0F,*/ |
1632 | /*0x1F18, 0x1F19, 0x1F1A, 0x1F1B, 0x1F1C, 0x1F1D, 0x1F28, 0x1F29, 0x1F2A, 0x1F2B, |
1633 | 0x1F2C, 0x1F2D, 0x1F2E, 0x1F2F, 0x1F38, 0x1F39, 0x1F3A, 0x1F3B, 0x1F3C, 0x1F3D, |
1634 | 0x1F3E, 0x1F3F,*/ 0x1F48, 0x1F49, 0x1F4A, 0x1F4B, 0x1F4C, 0x1F4D, 0x1F59, 0x1F5B, |
1635 | 0x1F5D, 0x1F5F, 0x1F68, 0x1F69, 0x1F6A, 0x1F6B, 0x1F6C, 0x1F6D, 0x1F6E, 0x1F6F, |
1636 | 0x1FBA, 0x1FBB, 0x1FC8, 0x1FC9, 0x1FCA, 0x1FCB, 0x1FDA, 0x1FDB, 0x1FF8, 0x1FF9, |
1637 | 0x1FEA, 0x1FEB, 0x1FFA, 0x1FFB, 0x1F88, 0x1F89, 0x1F8A, 0x1F8B, 0x1F8C, 0x1F8D, |
1638 | 0x1F8E, 0x1F8F, 0x1F98, 0x1F99, 0x1F9A, 0x1F9B, 0x1F9C, 0x1F9D, 0x1F9E, 0x1F9F, |
1639 | 0x1FA8, 0x1FA9, 0x1FAA, 0x1FAB, 0x1FAC, 0x1FAD, 0x1FAE, 0x1FAF, 0x1FB8, 0x1FB9, |
1640 | 0x1FBC, 0x0399, 0x1FCC, 0x1FD8, 0x1FD9, 0x1FE8, 0x1FE9, 0x1FEC, 0x1FFC /*0x2160, |
1641 | 0x2161, 0x2162, 0x2163, 0x2164, 0x2165, 0x2166, 0x2167, 0x2168, 0x2169, 0x216A, |
1642 | 0x216B, 0x216C, 0x216D, 0x216E, 0x216F 0x24B6, 0x24B7, 0x24B8, 0x24B9, 0x24BA, |
1643 | 0x24BB, 0x24BC, 0x24BD, 0x24BE, 0x24BF, 0x24C0, 0x24C1, 0x24C2, 0x24C3, 0x24C4, |
1644 | 0x24C5, 0x24C6, 0x24C7, 0x24C8, 0x24C9, 0x24CA, 0x24CB, 0x24CC, 0x24CD, 0x24CE, |
1645 | 0x24CF 0xFF21, 0xFF22, 0xFF23, 0xFF24, 0xFF25, 0xFF26, 0xFF27, 0xFF28, 0xFF29, |
1646 | 0xFF2A, 0xFF2B, 0xFF2C, 0xFF2D, 0xFF2E, 0xFF2F, 0xFF30, 0xFF31, 0xFF32, 0xFF33, |
1647 | 0xFF34, 0xFF35, 0xFF36, 0xFF37, 0xFF38, 0xFF39, 0xFF3A*/ |
1648 | }; |
1649 | |
1650 | |
1651 | // 12sep02 grandma : table driven inline function is 14x faster than original function, |
1652 | // Using first 100 movies of ATS, HashKey alone calls CharToUpper() 360,000 times. |
1653 | |
1654 | // Only entries 0-254 are used in these tables in unicharToUpper and unicharToLower |
1655 | // so entry 255 has been removed. Entry 255 is handled by the lowerCaseBase code. |
1656 | |
1657 | static const uint8_t tolower_map[] = |
1658 | { |
1659 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0-15 |
1660 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //16-31 |
1661 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //32-47 |
1662 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //48-63 |
1663 | 0x00, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, //64-79 |
1664 | 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, //80-95 |
1665 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //96-111 |
1666 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //112-127 |
1667 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //128-143 |
1668 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //144-159 |
1669 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //160-175 |
1670 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //176-191 |
1671 | 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, //192-207 |
1672 | 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, //208-223 |
1673 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //224-239 |
1674 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 //240-254 |
1675 | }; |
1676 | |
1677 | static const uint8_t toupper_map[] = |
1678 | { |
1679 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //0-15 |
1680 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //16-31 |
1681 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //32-47 |
1682 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //48-63 |
1683 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //64-79 |
1684 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //80-95 |
1685 | 0x00, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, //96-111 |
1686 | 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, //112-127 |
1687 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //128-143 |
1688 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //144-159 |
1689 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //160-175 |
1690 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //176-191 |
1691 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //192-207 |
1692 | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, //208-223 |
1693 | 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, //224-239 |
1694 | 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x00, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20, 0x20 //240-254 |
1695 | }; |
1696 | |
1697 | |
1698 | // WARNING: This is used by the core flash code. Any change to this utility, or the tables |
1699 | // it relies on, will break legacy Flash content. |
1700 | /*static*/ uint32_t String::unicharToUpper(uint32_t ch) |
1701 | { |
1702 | if (ch < 0xFF) |
1703 | return toupper_map[ch] ^ ch; |
1704 | |
1705 | // offset 0x1C60 |
1706 | /* |
1707 | if ( (ch>=0x2D00 && ch<=0x2D25) ) // Georgian |
1708 | { |
1709 | return (ch - 0x1C60); |
1710 | } |
1711 | */ |
1712 | |
1713 | // offset 80 |
1714 | if ( (ch>=0x0450 && ch<=0x045F) ) |
1715 | { |
1716 | return (ch - 0x50); |
1717 | } |
1718 | |
1719 | // offset 48 |
1720 | if ( (ch>=0x0561 && ch<=0x0586) ) |
1721 | { |
1722 | return (ch - 0x30); |
1723 | } |
1724 | |
1725 | // offset -32 |
1726 | if ( (ch>=0x03B1 && ch<=0x03CB) || |
1727 | (ch>=0x0430 && ch<=0x044F) || |
1728 | (ch>=0xFF41 && ch<=0xFF5A) ) |
1729 | { |
1730 | return (ch - 0x20); |
1731 | } |
1732 | |
1733 | // offset 26 |
1734 | if ( (ch>=0x24D0 && ch<=0x24E9) ) |
1735 | { |
1736 | return (ch - 0x1A); |
1737 | } |
1738 | |
1739 | // offset 16 |
1740 | if ( (ch>=0x2170 && ch<=0x217F) ) |
1741 | { |
1742 | return (ch - 0x10); |
1743 | } |
1744 | |
1745 | // offset +8(positive) |
1746 | if ( (ch>=0x1F00 && ch<=0X1F07) || |
1747 | (ch>=0x1F10 && ch<=0x1F15) || |
1748 | (ch>=0x1F20 && ch<=0x1F27) || |
1749 | (ch>=0x1F30 && ch<=0x1F37) ) |
1750 | { |
1751 | return (ch + 0x8); |
1752 | } |
1753 | |
1754 | // offset -1 |
1755 | if ( (ch>=0x0101 && ch<=0x0233) ) |
1756 | { |
1757 | if ( ((ch<=0x012F) && (ch&0x1)) || // odd only |
1758 | ((ch>=0x013A && ch<=0x0148) && !(ch&0x1)) || // even only |
1759 | ((ch>=0x014B && ch<=0x0177) && (ch&1)) || // odd only |
1760 | ((ch>=0x0201 && ch<=0x0233) && (ch&1) && ch!=0x0221) ) // odd only |
1761 | { |
1762 | return (ch - 1); |
1763 | } |
1764 | } |
1765 | |
1766 | if ( ch&1 ) // only looking for odd chars here |
1767 | { |
1768 | // GREEK |
1769 | if ( ch>=0x03D9 && ch<=0x03EF ) |
1770 | return (ch - 1); |
1771 | // CYRILLIC |
1772 | if ( ((ch>=0x0461 && ch<=0x04BF) && !(ch==0x0483 || ch==0x0485 || ch==0x487 || ch==0x0489)) || |
1773 | (ch>=0x04D1 && ch<=0x04F9) ) |
1774 | { |
1775 | if (ch!=0x0483 && ch!=0x0485 && ch!=0x487 && ch!=0x0489) |
1776 | return (ch - 1); |
1777 | } |
1778 | // LATIN_EXT_ADD |
1779 | if ( (ch>=0x1E01 && ch<=0x1E95) || (ch>=0x1EA1 && ch<=0x1EF9) ) |
1780 | return (ch - 1); |
1781 | } |
1782 | |
1783 | |
1784 | AvmAssert(sizeof(lowerCaseBase) == sizeof(upperCaseConversion))do { } while (0); |
1785 | |
1786 | uint32_t result = ch; |
1787 | // Do a binary search in lowerCaseBase for char |
1788 | int32_t lo = 0; |
1789 | int32_t hi = (sizeof(lowerCaseBase) / sizeof(lowerCaseBase[0])) - 1; |
1790 | |
1791 | while (lo <= hi) |
1792 | { |
1793 | int32_t pivot = (lo+hi)>>1; |
1794 | uint32_t testChar = lowerCaseBase[pivot]; |
1795 | |
1796 | if (ch == testChar) |
1797 | { |
1798 | // Use that index into lowerCaseConversion for a return value |
1799 | result = upperCaseConversion[pivot]; |
1800 | break; |
1801 | } |
1802 | else if (ch < testChar) |
1803 | { |
1804 | hi = pivot-1; |
1805 | } |
1806 | else |
1807 | { |
1808 | lo = pivot+1; |
1809 | } |
1810 | } |
1811 | |
1812 | return result; |
1813 | } |
1814 | |
1815 | static const wchar upperCaseBase[] = |
1816 | { |
1817 | /*0x0100, 0x0102, 0x0104, 0x0106, |
1818 | 0x0108, 0x010A, 0x010C, 0x010E, 0x0110, 0x0112, 0x0114, 0x0116, 0x0118, 0x011A, |
1819 | 0x011C, 0x011E, 0x0120, 0x0122, 0x0124, 0x0126, 0x0128, 0x012A, 0x012C, 0x012E,*/ |
1820 | 0x0130, 0x0132, 0x0134, 0x0136, /*0x0139, 0x013B, 0x013D, 0x013F, 0x0141, 0x0143, |
1821 | 0x0145, 0x0147, 0x014A, 0x014C, 0x014E, 0x0150, 0x0152, 0x0154, 0x0156, 0x0158, |
1822 | 0x015A, 0x015C, 0x015E, 0x0160, 0x0162, 0x0164, 0x0166, 0x0168, 0x016A, 0x016C, |
1823 | 0x016E, 0x0170, 0x0172, 0x0174, 0x0176,*/ 0x0178, 0x0179, 0x017B, 0x017D, 0x0181, |
1824 | 0x0182, 0x0184, 0x0186, 0x0187, 0x0189, 0x018A, 0x018B, 0x018E, 0x018F, 0x0190, |
1825 | 0x0191, 0x0193, 0x0194, 0x0196, 0x0197, 0x0198, 0x019C, 0x019D, 0x019F, 0x01A0, |
1826 | 0x01A2, 0x01A4, 0x01A6, 0x01A7, 0x01A9, 0x01AC, 0x01AE, 0x01AF, 0x01B1, 0x01B2, |
1827 | 0x01B3, 0x01B5, 0x01B7, 0x01B8, 0x01BC, 0x01C4, 0x01C5, 0x01C7, 0x01C8, 0x01CA, |
1828 | 0x01CB, 0x01CD, 0x01CF, 0x01D1, 0x01D3, 0x01D5, 0x01D7, 0x01D9, 0x01DB, 0x01DE, |
1829 | 0x01E0, 0x01E2, 0x01E4, 0x01E6, 0x01E8, 0x01EA, 0x01EC, 0x01EE, 0x01F1, 0x01F2, |
1830 | 0x01F4, 0x01F6, 0x01F7, 0x01F8, 0x01FA, 0x01FC, 0x01FE, /*0x0200, 0x0202, 0x0204, |
1831 | 0x0206, 0x0208, 0x020A, 0x020C, 0x020E, 0x0210, 0x0212, 0x0214, 0x0216, 0x0218, |
1832 | 0x021A, 0x021C, 0x021E, 0x0222, 0x0224, 0x0226, 0x0228, 0x022A, 0x022C, 0x022E, |
1833 | 0x0230, 0x0232,*/ 0x0386, 0x0388, 0x0389, 0x038A, 0x038C, 0x038E, 0x038F, /*0x0391, |
1834 | 0x0392, 0x0393, 0x0394, 0x0395, 0x0396, 0x0397, 0x0398, 0x0399, 0x039A, 0x039B, |
1835 | 0x039C, 0x039D, 0x039E, 0x039F, 0x03A0, 0x03A1, 0x03A3, 0x03A4, 0x03A5, 0x03A6, |
1836 | 0x03A7, 0x03A8, 0x03A9, 0x03AA, 0x03AB, 0x03DA, 0x03DC, 0x03DE, 0x03E0, 0x03E2, |
1837 | 0x03E4, 0x03E6, 0x03E8, 0x03EA, 0x03EC, 0x03EE,*/ 0x03F4, /*0x0400, 0x0401, 0x0402, |
1838 | 0x0403, 0x0404, 0x0405, 0x0406, 0x0407, 0x0408, 0x0409, 0x040A, 0x040B, 0x040C, |
1839 | 0x040D, 0x040E, 0x040F,*/ /*0x0410, 0x0411, 0x0412, 0x0413, 0x0414, 0x0415, 0x0416, |
1840 | 0x0417, 0x0418, 0x0419, 0x041A, 0x041B, 0x041C, 0x041D, 0x041E, 0x041F, 0x0420, |
1841 | 0x0421, 0x0422, 0x0423, 0x0424, 0x0425, 0x0426, 0x0427, 0x0428, 0x0429, 0x042A, |
1842 | 0x042B, 0x042C, 0x042D, 0x042E, 0x042F, 0x0460, 0x0462, 0x0464, 0x0466, 0x0468, |
1843 | 0x046A, 0x046C, 0x046E, 0x0470, 0x0472, 0x0474, 0x0476, 0x0478, 0x047A, 0x047C, |
1844 | 0x047E, 0x0480, 0x048C, 0x048E, 0x0490, 0x0492, 0x0494, 0x0496, 0x0498, 0x049A, |
1845 | 0x049C, 0x049E, 0x04A0, 0x04A2, 0x04A4, 0x04A6, 0x04A8, 0x04AA, 0x04AC, 0x04AE, |
1846 | 0x04B0, 0x04B2, 0x04B4, 0x04B6, 0x04B8, 0x04BA, 0x04BC, 0x04BE,*/ 0x04C1, 0x04C3, |
1847 | 0x04C7, 0x04CB, /*0x04D0, 0x04D2, 0x04D4, 0x04D6, 0x04D8, 0x04DA, 0x04DC, 0x04DE, |
1848 | 0x04E0, 0x04E2, 0x04E4, 0x04E6, 0x04E8, 0x04EA, 0x04EC, 0x04EE, 0x04F0, 0x04F2, |
1849 | 0x04F4, 0x04F8, 0x0531, 0x0532, 0x0533, 0x0534, 0x0535, 0x0536, 0x0537, 0x0538, |
1850 | 0x0539, 0x053A, 0x053B, 0x053C, 0x053D, 0x053E, 0x053F, 0x0540, 0x0541, 0x0542, |
1851 | 0x0543, 0x0544, 0x0545, 0x0546, 0x0547, 0x0548, 0x0549, 0x054A, 0x054B, 0x054C, |
1852 | 0x054D, 0x054E, 0x054F, 0x0550, 0x0551, 0x0552, 0x0553, 0x0554, 0x0555, 0x0556,*/ |
1853 | /* |
1854 | // cn: added Georgian. Not in the UnicodeData-3.2.0 spreadsheet, but was added later |
1855 | 0x10a0, 0x10a1, 0x10a2, 0x10a3, 0x10a4, 0x10a5, 0x10a6, 0x10a7, 0x10a8, 0x10a9, |
1856 | 0x10aa, 0x10ab, 0x10ac, 0x10ad, 0x10ae, 0x10af, 0x10b0, 0x10b1, 0x10b2, 0x10b3, |
1857 | 0x10b4, 0x10b5, 0x10b6, 0x10b7, 0x10b8, 0x10b9, 0x10ba, 0x10bb, 0x10bc, 0x10bd, |
1858 | 0x10be, 0x10bf, 0x10c0, 0x10c1, 0x10c2, 0x10c3, 0x10c4, 0x10c5, |
1859 | // cn: end Georgian. |
1860 | */ |
1861 | /*0x1E00, 0x1E02, 0x1E04, 0x1E06, 0x1E08, 0x1E0A, 0x1E0C, 0x1E0E, 0x1E10, 0x1E12, |
1862 | 0x1E14, 0x1E16, 0x1E18, 0x1E1A, 0x1E1C, 0x1E1E, 0x1E20, 0x1E22, 0x1E24, 0x1E26, |
1863 | 0x1E28, 0x1E2A, 0x1E2C, 0x1E2E, 0x1E30, 0x1E32, 0x1E34, 0x1E36, 0x1E38, 0x1E3A, |
1864 | 0x1E3C, 0x1E3E, 0x1E40, 0x1E42, 0x1E44, 0x1E46, 0x1E48, 0x1E4A, 0x1E4C, 0x1E4E, |
1865 | 0x1E50, 0x1E52, 0x1E54, 0x1E56, 0x1E58, 0x1E5A, 0x1E5C, 0x1E5E, 0x1E60, 0x1E62, |
1866 | 0x1E64, 0x1E66, 0x1E68, 0x1E6A, 0x1E6C, 0x1E6E, 0x1E70, 0x1E72, 0x1E74, 0x1E76, |
1867 | 0x1E78, 0x1E7A, 0x1E7C, 0x1E7E, 0x1E80, 0x1E82, 0x1E84, 0x1E86, 0x1E88, 0x1E8A, |
1868 | 0x1E8C, 0x1E8E, 0x1E90, 0x1E92, 0x1E94, 0x1EA0, 0x1EA2, 0x1EA4, 0x1EA6, 0x1EA8, |
1869 | 0x1EAA, 0x1EAC, 0x1EAE, 0x1EB0, 0x1EB2, 0x1EB4, 0x1EB6, 0x1EB8, 0x1EBA, 0x1EBC, |
1870 | 0x1EBE, 0x1EC0, 0x1EC2, 0x1EC4, 0x1EC6, 0x1EC8, 0x1ECA, 0x1ECC, 0x1ECE, 0x1ED0, |
1871 | 0x1ED2, 0x1ED4, 0x1ED6, 0x1ED8, 0x1EDA, 0x1EDC, 0x1EDE, 0x1EE0, 0x1EE2, 0x1EE4, |
1872 | 0x1EE6, 0x1EE8, 0x1EEA, 0x1EEC, 0x1EEE, 0x1EF0, 0x1EF2, 0x1EF4, 0x1EF6, 0x1EF8, |
1873 | 0x1F08, 0x1F09, 0x1F0A, 0x1F0B, 0x1F0C, 0x1F0D, 0x1F0E, 0x1F0F, 0x1F18, 0x1F19, |
1874 | 0x1F1A, 0x1F1B, 0x1F1C, 0x1F1D, 0x1F28, 0x1F29, 0x1F2A, 0x1F2B, 0x1F2C, 0x1F2D, |
1875 | 0x1F2E, 0x1F2F, 0x1F38, 0x1F39, 0x1F3A, 0x1F3B, 0x1F3C, 0x1F3D, 0x1F3E, 0x1F3F,*/ |
1876 | 0x1F48, 0x1F49, 0x1F4A, 0x1F4B, 0x1F4C, 0x1F4D, 0x1F59, 0x1F5B, 0x1F5D, 0x1F5F, |
1877 | 0x1F68, 0x1F69, 0x1F6A, 0x1F6B, 0x1F6C, 0x1F6D, 0x1F6E, 0x1F6F, 0x1F88, 0x1F89, |
1878 | 0x1F8A, 0x1F8B, 0x1F8C, 0x1F8D, 0x1F8E, 0x1F8F, 0x1F98, 0x1F99, 0x1F9A, 0x1F9B, |
1879 | 0x1F9C, 0x1F9D, 0x1F9E, 0x1F9F, 0x1FA8, 0x1FA9, 0x1FAA, 0x1FAB, 0x1FAC, 0x1FAD, |
1880 | 0x1FAE, 0x1FAF, 0x1FB8, 0x1FB9, 0x1FBA, 0x1FBB, 0x1FBC, 0x1FC8, 0x1FC9, 0x1FCA, |
1881 | 0x1FCB, 0x1FCC, 0x1FD8, 0x1FD9, 0x1FDA, 0x1FDB, 0x1FE8, 0x1FE9, 0x1FEA, 0x1FEB, |
1882 | 0x1FEC, 0x1FF8, 0x1FF9, 0x1FFA, 0x1FFB, 0x1FFC, 0x2126, 0x212A, 0x212B /*0x2160, |
1883 | 0x2161, 0x2162, 0x2163, 0x2164, 0x2165, 0x2166, 0x2167, 0x2168, 0x2169, 0x216A, |
1884 | 0x216B, 0x216C, 0x216D, 0x216E, 0x216F, 0x24B6, 0x24B7, 0x24B8, 0x24B9, 0x24BA, |
1885 | 0x24BB, 0x24BC, 0x24BD, 0x24BE, 0x24BF, 0x24C0, 0x24C1, 0x24C2, 0x24C3, 0x24C4, |
1886 | 0x24C5, 0x24C6, 0x24C7, 0x24C8, 0x24C9, 0x24CA, 0x24CB, 0x24CC, 0x24CD, 0x24CE, |
1887 | 0x24CF, 0xFF21, 0xFF22, 0xFF23, 0xFF24, 0xFF25, 0xFF26, 0xFF27, 0xFF28, 0xFF29, |
1888 | 0xFF2A, 0xFF2B, 0xFF2C, 0xFF2D, 0xFF2E, 0xFF2F, 0xFF30, 0xFF31, 0xFF32, 0xFF33, |
1889 | 0xFF34, 0xFF35, 0xFF36, 0xFF37, 0xFF38, 0xFF39, 0xFF3A*/ |
1890 | }; |
1891 | |
1892 | static const wchar lowerCaseConversion[] = |
1893 | { |
1894 | /*0x0101, 0x0103, 0x0105, 0x0107, |
1895 | 0x0109, 0x010B, 0x010D, 0x010F, 0x0111, 0x0113, 0x0115, 0x0117, 0x0119, 0x011B, |
1896 | 0x011D, 0x011F, 0x0121, 0x0123, 0x0125, 0x0127, 0x0129, 0x012B, 0x012D, 0x012F,*/ |
1897 | 0x0069, 0x0133, 0x0135, 0x0137, /*0x013A, 0x013C, 0x013E, 0x0140, 0x0142, 0x0144, |
1898 | 0x0146, 0x0148, 0x014B, 0x014D, 0x014F, 0x0151, 0x0153, 0x0155, 0x0157, 0x0159, |
1899 | 0x015B, 0x015D, 0x015F, 0x0161, 0x0163, 0x0165, 0x0167, 0x0169, 0x016B, 0x016D, |
1900 | 0x016F, 0x0171, 0x0173, 0x0175, 0x0177,*/ 0x00FF, 0x017A, 0x017C, 0x017E, 0x0253, |
1901 | 0x0183, 0x0185, 0x0254, 0x0188, 0x0256, 0x0257, 0x018C, 0x01DD, 0x0259, 0x025B, |
1902 | 0x0192, 0x0260, 0x0263, 0x0269, 0x0268, 0x0199, 0x026F, 0x0272, 0x0275, 0x01A1, |
1903 | 0x01A3, 0x01A5, 0x0280, 0x01A8, 0x0283, 0x01AD, 0x0288, 0x01B0, 0x028A, 0x028B, |
1904 | 0x01B4, 0x01B6, 0x0292, 0x01B9, 0x01BD, 0x01C6, 0x01C6, 0x01C9, 0x01C9, 0x01CC, |
1905 | 0x01CC, 0x01CE, 0x01D0, 0x01D2, 0x01D4, 0x01D6, 0x01D8, 0x01DA, 0x01DC, 0x01DF, |
1906 | 0x01E1, 0x01E3, 0x01E5, 0x01E7, 0x01E9, 0x01EB, 0x01ED, 0x01EF, 0x01F3, 0x01F3, |
1907 | 0x01F5, 0x0195, 0x01BF, 0x01F9, 0x01FB, 0x01FD, 0x01FF, /*0x0201, 0x0203, 0x0205, |
1908 | 0x0207, 0x0209, 0x020B, 0x020D, 0x020F, 0x0211, 0x0213, 0x0215, 0x0217, 0x0219, |
1909 | 0x021B, 0x021D, 0x021F, 0x0223, 0x0225, 0x0227, 0x0229, 0x022B, 0x022D, 0x022F, |
1910 | 0x0231, 0x0233,*/ 0x03AC, 0x03AD, 0x03AE, 0x03AF, 0x03CC, 0x03CD, 0x03CE, /*0x03B1, |
1911 | 0x03B2, 0x03B3, 0x03B4, 0x03B5, 0x03B6, 0x03B7, 0x03B8, 0x03B9, 0x03BA, 0x03BB, |
1912 | 0x03BC, 0x03BD, 0x03BE, 0x03BF, 0x03C0, 0x03C1, 0x03C3, 0x03C4, 0x03C5, 0x03C6, |
1913 | 0x03C7, 0x03C8, 0x03C9, 0x03CA, 0x03CB, 0x03DB, 0x03DD, 0x03DF, 0x03E1, 0x03E3, |
1914 | 0x03E5, 0x03E7, 0x03E9, 0x03EB, 0x03ED, 0x03EF,*/ 0x03B8, /*0x0450, 0x0451, 0x0452, |
1915 | 0x0453, 0x0454, 0x0455, 0x0456, 0x0457, 0x0458, 0x0459, 0x045A, 0x045B, 0x045C, |
1916 | 0x045D, 0x045E, 0x045F,*/ /*0x0430, 0x0431, 0x0432, 0x0433, 0x0434, 0x0435, 0x0436, |
1917 | 0x0437, 0x0438, 0x0439, 0x043A, 0x043B, 0x043C, 0x043D, 0x043E, 0x043F, 0x0440, |
1918 | 0x0441, 0x0442, 0x0443, 0x0444, 0x0445, 0x0446, 0x0447, 0x0448, 0x0449, 0x044A, |
1919 | 0x044B, 0x044C, 0x044D, 0x044E, 0x044F, 0x0461, 0x0463, 0x0465, 0x0467, 0x0469, |
1920 | 0x046B, 0x046D, 0x046F, 0x0471, 0x0473, 0x0475, 0x0477, 0x0479, 0x047B, 0x047D, |
1921 | 0x047F, 0x0481, 0x048D, 0x048F, 0x0491, 0x0493, 0x0495, 0x0497, 0x0499, 0x049B, |
1922 | 0x049D, 0x049F, 0x04A1, 0x04A3, 0x04A5, 0x04A7, 0x04A9, 0x04AB, 0x04AD, 0x04AF, |
1923 | 0x04B1, 0x04B3, 0x04B5, 0x04B7, 0x04B9, 0x04BB, 0x04BD, 0x04BF,*/ 0x04C2, 0x04C4, |
1924 | 0x04C8, 0x04CC, /*0x04D1, 0x04D3, 0x04D5, 0x04D7, 0x04D9, 0x04DB, 0x04DD, 0x04DF, |
1925 | 0x04E1, 0x04E3, 0x04E5, 0x04E7, 0x04E9, 0x04EB, 0x04ED, 0x04EF, 0x04F1, 0x04F3, |
1926 | 0x04F5, 0x04F9, 0x0561, 0x0562, 0x0563, 0x0564, 0x0565, 0x0566, 0x0567, 0x0568, |
1927 | 0x0569, 0x056A, 0x056B, 0x056C, 0x056D, 0x056E, 0x056F, 0x0570, 0x0571, 0x0572, |
1928 | 0x0573, 0x0574, 0x0575, 0x0576, 0x0577, 0x0578, 0x0579, 0x057A, 0x057B, 0x057C, |
1929 | 0x057D, 0x057E, 0x057F, 0x0580, 0x0581, 0x0582, 0x0583, 0x0584, 0x0585, 0x0586,*/ |
1930 | // cn: added Georgian. Not in the UnicodeData-3.2.0 spreadsheet, but was added later |
1931 | /* |
1932 | 0x10d0, 0x10d1, 0x10d2, 0x10d3, 0x10d4, 0x10d5, 0x10d6, 0x10d7, 0x10d8, 0x10d9, |
1933 | 0x10da, 0x10db, 0x10dc, 0x10dd, 0x10de, 0x10df, 0x10e0, 0x10e1, 0x10e2, 0x10e3, |
1934 | 0x10e4, 0x10e5, 0x10e6, 0x10e7, 0x10e8, 0x10e9, 0x10ea, 0x10eb, 0x10ec, 0x10ed, |
1935 | 0x10ee, 0x10ef, 0x10f0, 0x10f1, 0x10f2, 0x10f3, 0x10f4, 0x10f5, |
1936 | // cn: end Georgian. |
1937 | */ |
1938 | /*0x1E01, 0x1E03, 0x1E05, 0x1E07, 0x1E09, 0x1E0B, 0x1E0D, 0x1E0F, 0x1E11, 0x1E13, |
1939 | 0x1E15, 0x1E17, 0x1E19, 0x1E1B, 0x1E1D, 0x1E1F, 0x1E21, 0x1E23, 0x1E25, 0x1E27, |
1940 | 0x1E29, 0x1E2B, 0x1E2D, 0x1E2F, 0x1E31, 0x1E33, 0x1E35, 0x1E37, 0x1E39, 0x1E3B, |
1941 | 0x1E3D, 0x1E3F, 0x1E41, 0x1E43, 0x1E45, 0x1E47, 0x1E49, 0x1E4B, 0x1E4D, 0x1E4F, |
1942 | 0x1E51, 0x1E53, 0x1E55, 0x1E57, 0x1E59, 0x1E5B, 0x1E5D, 0x1E5F, 0x1E61, 0x1E63, |
1943 | 0x1E65, 0x1E67, 0x1E69, 0x1E6B, 0x1E6D, 0x1E6F, 0x1E71, 0x1E73, 0x1E75, 0x1E77, |
1944 | 0x1E79, 0x1E7B, 0x1E7D, 0x1E7F, 0x1E81, 0x1E83, 0x1E85, 0x1E87, 0x1E89, 0x1E8B, |
1945 | 0x1E8D, 0x1E8F, 0x1E91, 0x1E93, 0x1E95, 0x1EA1, 0x1EA3, 0x1EA5, 0x1EA7, 0x1EA9, |
1946 | 0x1EAB, 0x1EAD, 0x1EAF, 0x1EB1, 0x1EB3, 0x1EB5, 0x1EB7, 0x1EB9, 0x1EBB, 0x1EBD, |
1947 | 0x1EBF, 0x1EC1, 0x1EC3, 0x1EC5, 0x1EC7, 0x1EC9, 0x1ECB, 0x1ECD, 0x1ECF, 0x1ED1, |
1948 | 0x1ED3, 0x1ED5, 0x1ED7, 0x1ED9, 0x1EDB, 0x1EDD, 0x1EDF, 0x1EE1, 0x1EE3, 0x1EE5, |
1949 | 0x1EE7, 0x1EE9, 0x1EEB, 0x1EED, 0x1EEF, 0x1EF1, 0x1EF3, 0x1EF5, 0x1EF7, 0x1EF9, |
1950 | 0x1F00, 0x1F01, 0x1F02, 0x1F03, 0x1F04, 0x1F05, 0x1F06, 0x1F07, 0x1F10, 0x1F11, |
1951 | 0x1F12, 0x1F13, 0x1F14, 0x1F15, 0x1F20, 0x1F21, 0x1F22, 0x1F23, 0x1F24, 0x1F25, |
1952 | 0x1F26, 0x1F27, 0x1F30, 0x1F31, 0x1F32, 0x1F33, 0x1F34, 0x1F35, 0x1F36, 0x1F37,*/ |
1953 | 0x1F40, 0x1F41, 0x1F42, 0x1F43, 0x1F44, 0x1F45, 0x1F51, 0x1F53, 0x1F55, 0x1F57, |
1954 | 0x1F60, 0x1F61, 0x1F62, 0x1F63, 0x1F64, 0x1F65, 0x1F66, 0x1F67, 0x1F80, 0x1F81, |
1955 | 0x1F82, 0x1F83, 0x1F84, 0x1F85, 0x1F86, 0x1F87, 0x1F90, 0x1F91, 0x1F92, 0x1F93, |
1956 | 0x1F94, 0x1F95, 0x1F96, 0x1F97, 0x1FA0, 0x1FA1, 0x1FA2, 0x1FA3, 0x1FA4, 0x1FA5, |
1957 | 0x1FA6, 0x1FA7, 0x1FB0, 0x1FB1, 0x1F70, 0x1F71, 0x1FB3, 0x1F72, 0x1F73, 0x1F74, |
1958 | 0x1F75, 0x1FC3, 0x1FD0, 0x1FD1, 0x1F76, 0x1F77, 0x1FE0, 0x1FE1, 0x1F7A, 0x1F7B, |
1959 | 0x1FE5, 0x1F78, 0x1F79, 0x1F7C, 0x1F7D, 0x1FF3, 0x03C9, 0x006B, 0x00E5 /*0x2170, |
1960 | 0x2171, 0x2172, 0x2173, 0x2174, 0x2175, 0x2176, 0x2177, 0x2178, 0x2179, 0x217A, |
1961 | 0x217B, 0x217C, 0x217D, 0x217E, 0x217F, 0x24D0, 0x24D1, 0x24D2, 0x24D3, 0x24D4, |
1962 | 0x24D5, 0x24D6, 0x24D7, 0x24D8, 0x24D9, 0x24DA, 0x24DB, 0x24DC, 0x24DD, 0x24DE, |
1963 | 0x24DF, 0x24E0, 0x24E1, 0x24E2, 0x24E3, 0x24E4, 0x24E5, 0x24E6, 0x24E7, 0x24E8, |
1964 | 0x24E9, 0xFF41, 0xFF42, 0xFF43, 0xFF44, 0xFF45, 0xFF46, 0xFF47, 0xFF48, 0xFF49, |
1965 | 0xFF4A, 0xFF4B, 0xFF4C, 0xFF4D, 0xFF4E, 0xFF4F, 0xFF50, 0xFF51, 0xFF52, 0xFF53, |
1966 | 0xFF54, 0xFF55, 0xFF56, 0xFF57, 0xFF58, 0xFF59, 0xFF5A*/ |
1967 | }; |
1968 | |
1969 | |
1970 | // WARNING: This is used by the core flash code. Any change to this utility, or the tables |
1971 | // it relies on, will break legacy Flash content. |
1972 | /*static*/ uint32_t String::unicharToLower(uint32_t ch) |
1973 | { |
1974 | if (ch < 0xFF) |
1975 | return tolower_map[ch] ^ ch; |
1976 | |
1977 | // offset x1C60 |
1978 | if( ch>=0x10A0 && ch<=0x10C5 ) // Georgian |
1979 | { |
1980 | return (ch + 48); // WRONG |
1981 | //return (ch + 0x1C60); // CORRECT |
1982 | } |
1983 | |
1984 | // offset 80 |
1985 | if ( (ch>=0x0400 && ch<=0x040F) ) |
1986 | { |
1987 | return (ch + 0x50); |
1988 | } |
1989 | |
1990 | // offset 48 |
1991 | if ( (ch>=0x0531 && ch<=0x0556) ) |
1992 | { |
1993 | return (ch + 0x30); |
1994 | } |
1995 | |
1996 | // offset 32 |
1997 | if ( (ch>=0x0391 && ch<=0x03AB) || |
1998 | (ch>=0x0410 && ch<=0x042F) || |
1999 | (ch>=0xFF21 && ch<=0xFF3A) ) |
2000 | { |
2001 | return (ch + 0x20); |
2002 | } |
2003 | |
2004 | // offset 26 |
2005 | if ( (ch>=0x24B6 && ch<=0x24CF) ) |
2006 | { |
2007 | return (ch + 0x1A); |
2008 | } |
2009 | |
2010 | // offset 16 |
2011 | if ( (ch>=0x2160 && ch<=0x216F) ) |
2012 | { |
2013 | return (ch + 0x10); |
2014 | } |
2015 | |
2016 | // offset -8(negative) |
2017 | if ( (ch>=0x1F08 && ch<=0X1F0F) || |
2018 | (ch>=0x1F18 && ch<=0x1F1D) || |
2019 | (ch>=0x1F28 && ch<=0x1F2F) || |
2020 | (ch>=0x1F38 && ch<=0x1F3F) ) |
2021 | { |
2022 | return (ch - 0x8); |
2023 | } |
2024 | |
2025 | // offset +1 |
2026 | if ( (ch>=0x0100 && ch<=0x0232) ) |
2027 | { |
2028 | // only even numbers |
2029 | if ( ((ch<=0x012E) && !(ch&0x1)) || |
2030 | ((ch>=0x0139 && ch<=0x0147) && (ch&0x1)) || |
2031 | ((ch>=0x014A && ch<=0x0176) && !(ch&1)) || |
2032 | ((ch>=0x0200 && ch<=0x0232) && !(ch&1) && (ch!=0x0220)) ) |
2033 | { |
2034 | return (ch + 1); |
2035 | } |
2036 | } |
2037 | |
2038 | if ( !(ch&1) ) // only looking for even chars |
2039 | { |
2040 | if ( (ch>=0x03D8 && ch<=0x03EE) ) |
2041 | return (ch + 1); |
2042 | if ( ((ch>=0x0460 && ch<=0x04BE) && !(ch==0x0482 || ch==0x0484 || ch==0x486 || ch==0x0488)) || |
2043 | (ch>=0x04D0 && ch<=0x04F8) ) |
2044 | { |
2045 | if (ch!=0x0482 && ch!=0x0484 && ch!=0x486 && ch!=0x0488) |
2046 | return (ch + 1); |
2047 | } |
2048 | if ( (ch>=0x1E00 && ch<=0x1E94) || (ch>=0x1EA0 && ch<=0x1EF8) ) |
2049 | return (ch + 1); |
2050 | } |
2051 | |
2052 | |
2053 | AvmAssert(sizeof(upperCaseBase) == sizeof(lowerCaseConversion))do { } while (0); |
2054 | |
2055 | uint32_t result = ch; |
2056 | // Do a binary search in upperCaseBase for char |
2057 | int32_t lo = 0; |
2058 | int32_t hi = (sizeof(upperCaseBase) / sizeof(upperCaseBase[0])) - 1; |
2059 | |
2060 | while (lo <= hi) |
2061 | { |
2062 | int32_t pivot = (lo+hi)>>1; |
2063 | uint32_t testChar = upperCaseBase[pivot]; |
2064 | |
2065 | if (ch == testChar) |
2066 | { |
2067 | // Use that index into lowerCaseConversion for a return value |
2068 | result = lowerCaseConversion[pivot]; |
2069 | break; |
2070 | } |
2071 | else if (ch < testChar) |
2072 | { |
2073 | hi = pivot-1; |
2074 | } |
2075 | else |
2076 | { |
2077 | lo = pivot+1; |
2078 | } |
2079 | } |
2080 | |
2081 | return result; |
2082 | } |
2083 | |
2084 | Stringp String::toUpperCase() |
2085 | { |
2086 | return caseChange(unicharToUpper); |
2087 | } |
2088 | |
2089 | Stringp String::toLowerCase() |
2090 | { |
2091 | return caseChange(unicharToLower); |
2092 | } |
2093 | |
2094 | Stringp String::caseChange(uint32_t(*unimapper)(uint32_t)) |
2095 | { |
2096 | if (!this->length()) |
2097 | return this; |
2098 | |
2099 | // Flag to detect whether any changes were made |
2100 | boolbool changed = falsefalse; |
2101 | |
2102 | // 0xFF is a special case: ToUpper(0xFF) == 0x178, so we need a wider string |
2103 | // if the string contains 0xFF |
2104 | Width w = getWidth(); |
2105 | if (w == k8 && VMPI_memchr::memchr(Pointers(this).p8, 0xFF, m_length) != 0) |
2106 | w = k16; |
2107 | |
2108 | GC* gc = _gc(this); |
2109 | const boolbool is7bit = falsefalse; |
2110 | Stringp newStr = createDynamic(gc, NULL__null, m_length, w, is7bit); |
2111 | |
2112 | Pointers src(this); // can't re-use the Pointers from VMPI_memchr; createDynamic may have invalidated it |
2113 | Pointers dst(newStr); |
2114 | |
2115 | int32_t i; |
2116 | uint32_t ch1, ch2; |
2117 | if (getWidth() == k8) |
2118 | { |
2119 | for (i = 0; i < m_length; i++) |
2120 | { |
2121 | ch1 = uint32_t(*src.p8++); |
2122 | ch2 = unimapper (ch1); |
2123 | if (w == k16) |
2124 | *dst.p16++ = wchar(ch2); |
2125 | else |
2126 | *dst.p8++ = uint8_t(ch2); |
2127 | if (ch1 != ch2) |
2128 | changed = truetrue; |
2129 | } |
2130 | } |
2131 | else |
2132 | { |
2133 | for (i = 0; i < m_length; i++) |
2134 | { |
2135 | ch1 = uint32_t(*src.p16++); |
2136 | ch2 = unimapper(ch1); |
2137 | *dst.p16++ = wchar(ch2); |
2138 | if (ch1 != ch2) |
2139 | changed = truetrue; |
2140 | } |
2141 | } |
2142 | VERIFY_7BIT(this)do { } while (0); |
2143 | VERIFY_7BIT(newStr)do { } while (0); |
2144 | return changed ? newStr : this; |
2145 | } |
2146 | |
2147 | double String::toNumber() |
2148 | { |
2149 | return MathUtils::convertStringToNumber(this); |
2150 | } |
2151 | |
2152 | boolbool String::isWhitespace() const |
2153 | { |
2154 | StringIndexer self((Stringp) this); |
2155 | for (int32_t i = 0 ; i < length(); i++) |
2156 | { |
2157 | wchar ch = self[i]; |
2158 | if (!isSpace(ch)) |
2159 | return falsefalse; |
2160 | } |
2161 | return truetrue; |
2162 | } |
2163 | |
2164 | /*static*/ int32_t FASTCALL__attribute__((fastcall)) String::Length(const wchar* str) |
2165 | { |
2166 | AvmAssert(str != NULL)do { } while (0); |
2167 | |
2168 | const wchar* s = str; |
2169 | while (*s) |
2170 | s++; |
2171 | ptrdiff_t len = s - str; |
2172 | if (len <= 0x7FFFFFFF) |
2173 | return int32_t(len); |
2174 | GCHeap::SignalObjectTooLarge(); |
2175 | /*NOTREACHED*/ |
2176 | return 0; |
2177 | } |
2178 | |
2179 | /*static*/ int32_t FASTCALL__attribute__((fastcall)) String::Length(const char* str) |
2180 | { |
2181 | AvmAssert(str != NULL)do { } while (0); |
2182 | |
2183 | size_t len = VMPI_strlen::strlen(str); |
2184 | if (len <= 0x7FFFFFFF) |
2185 | return int32_t(len); |
2186 | GCHeap::SignalObjectTooLarge(); |
2187 | /*NOTREACHED*/ |
2188 | return 0; |
2189 | } |
2190 | |
2191 | int String::_indexOf(Stringp substr, int startPos) |
2192 | { |
2193 | return (int) indexOf(substr, (int32_t) startPos); |
2194 | } |
2195 | |
2196 | // Coded as a macro for best performance with MSVC 2008 32-bit |
2197 | // -1 (or negative infinity) - position equals -1 (out of bounds) |
2198 | // >length (or positive infinity) - position equals length |
2199 | // non-NaN - integer cast to determine position |
2200 | // NaN - use NaNValue (usually zero but length for lastIndexOf) |
2201 | #define INTCLAMP(iPos, dPos, length, NaNValue){ if (dPos <= -1) iPos = -1; else if (dPos >= length) iPos = length; else if (!MathUtils::isNaNInline(dPos)) { iPos = ( int32_t) dPos; } else { iPos = NaNValue; } } \ |
2202 | { \ |
2203 | if (dPos <= -1) \ |
2204 | iPos = -1; \ |
2205 | else if (dPos >= length) \ |
2206 | iPos = length; \ |
2207 | else if (!MathUtils::isNaNInline(dPos)) { \ |
2208 | iPos = (int32_t) dPos; \ |
2209 | } \ |
2210 | else { \ |
2211 | iPos = NaNValue; \ |
2212 | } \ |
2213 | } |
2214 | |
2215 | int32_t String::AS3_indexOf(Stringp substr, double dStartPos) |
2216 | { |
2217 | int32_t iPos; |
2218 | INTCLAMP(iPos, dStartPos, this->length(), 0){ if (dStartPos <= -1) iPos = -1; else if (dStartPos >= this->length()) iPos = this->length(); else if (!MathUtils ::isNaNInline(dStartPos)) { iPos = (int32_t) dStartPos; } else { iPos = 0; } }; |
2219 | return indexOf(substr, iPos); |
2220 | } |
2221 | |
2222 | int String::_lastIndexOf(Stringp substr, int iStartPos) |
2223 | { |
2224 | return (int) lastIndexOf(substr, (int32_t) iStartPos); |
2225 | } |
2226 | |
2227 | int32_t String::AS3_lastIndexOf(Stringp substr, double dStartPos) |
2228 | { |
2229 | // unlike most other calls, this one has nan->length rather than nan->0 |
2230 | int32_t iStartPos; |
2231 | INTCLAMP(iStartPos, dStartPos, this->length(), this->length()){ if (dStartPos <= -1) iStartPos = -1; else if (dStartPos >= this->length()) iStartPos = this->length(); else if (! MathUtils::isNaNInline(dStartPos)) { iStartPos = (int32_t) dStartPos ; } else { iStartPos = this->length(); } }; |
2232 | return lastIndexOf(substr, iStartPos); |
2233 | } |
2234 | |
2235 | Stringp String::_substr(int start, int count) |
2236 | { |
2237 | int len = this->length(); |
2238 | start = (int)NativeObjectHelpers::ClampIndexInt(start, len); |
2239 | // ClampIndex takes a double (not int or uint) for first parm... |
2240 | // we must cast these to double before addition, otherwise we |
2241 | // can have numeric overflow with the default arg (end=0x7fffffff) |
2242 | // and wrap to negative, which would be bad... |
2243 | |
2244 | // Do some sanity checks on our ints to see if they will fall within a valid integer range |
2245 | // !!@what about negative values? |
2246 | int end; |
2247 | if (count == 0x7fffffff) // largest positive int value, also the default arg |
2248 | { |
2249 | end = len; |
2250 | } |
2251 | else if ((count > 0x3fffffff) || (start > 0x3fffffff)) // might overflow - use doubles |
2252 | { |
2253 | end = (int)NativeObjectHelpers::ClampIndex(double(count) + double(start), len); |
2254 | } |
2255 | else |
2256 | { |
2257 | end = (int)NativeObjectHelpers::ClampIndexInt(count + start, len); |
2258 | } |
2259 | |
2260 | if (end < start) |
2261 | end = start; |
2262 | |
2263 | return substr(start, end-start); |
2264 | } |
2265 | |
2266 | Stringp String::AS3_substr(double d_start, double d_count) |
2267 | { |
2268 | int32_t len = this->length(); |
2269 | int32_t start = (int32_t) NativeObjectHelpers::ClampIndex(MathUtils::toInt(d_start), len); |
2270 | // ClampIndex takes a double (not int or uint) for first parm... |
2271 | // we must cast these to double before addition, otherwise we |
2272 | // can have numeric overflow with the default arg (end=0x7fffffff) |
2273 | // and wrap to negative, which would be bad... |
2274 | int32_t end = (int32_t)NativeObjectHelpers::ClampIndex(MathUtils::toInt(d_count) + (double)start, len); |
2275 | if (end < start) |
2276 | end = start; |
2277 | |
2278 | return substr(start, end-start); |
2279 | } |
2280 | |
2281 | Stringp String::_substring(int start, int end) |
2282 | { |
2283 | NativeObjectHelpers::ClampBInt(start, end, this->length()); |
2284 | return substring((int32_t) start, (int32_t) end); |
2285 | } |
2286 | |
2287 | Stringp String::AS3_substring(double d_start, double d_end) |
2288 | { |
2289 | double start = MathUtils::toInt(d_start); |
2290 | double end = MathUtils::toInt(d_end); |
2291 | NativeObjectHelpers::ClampB(start, end, length()); |
2292 | return substring((int32_t) start, (int32_t) end); |
2293 | } |
2294 | |
2295 | Stringp String::_slice(int start, int end) |
2296 | { |
2297 | return slice((int32_t) start, (int32_t) end); |
2298 | } |
2299 | |
2300 | Stringp String::AS3_slice(double d_start, double d_end) |
2301 | { |
2302 | int32_t len = this->length(); |
2303 | int32_t start = (int32_t)NativeObjectHelpers::ClampIndex(MathUtils::toInt(d_start), len); |
2304 | int32_t end = (int32_t)NativeObjectHelpers::ClampIndex(MathUtils::toInt(d_end), len); |
2305 | if (end < start) |
2306 | end = start; |
2307 | |
2308 | return substring(start, end); |
2309 | } |
2310 | |
2311 | Stringp String::_charAtI(int32_t iPos) |
2312 | { |
2313 | AvmCore* core = _core(this); |
2314 | // use unsigned compare as faster equivalent to >= 0 && < m_length |
2315 | if (uint32_t(iPos) < uint32_t(m_length)) { |
2316 | Pointers ptrs(this); |
2317 | wchar const ch = (wchar) (getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]; |
2318 | |
2319 | // newStringUTF16 does the cachedChar optimization internally, but short-circuiting the test |
2320 | // here is worthwhile because it's a pretty common case |
2321 | if (ch < 128) |
2322 | return core->cachedChars[ch]; |
2323 | return core->newStringUTF16(&ch, 1); |
2324 | } |
2325 | else { |
2326 | return core->kEmptyString; |
2327 | } |
2328 | } |
2329 | |
2330 | Stringp String::_charAtU(uint32_t iPos) |
2331 | { |
2332 | AvmCore* core = _core(this); |
2333 | if (iPos < uint32_t(m_length)) { |
2334 | Pointers ptrs(this); |
2335 | wchar const ch = (wchar) (getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]; |
2336 | |
2337 | // newStringUTF16 does the cachedChar optimization internally, but short-circuiting the test |
2338 | // here is worthwhile because it's a pretty common case |
2339 | if (ch < 128) |
2340 | return core->cachedChars[ch]; |
2341 | return core->newStringUTF16(&ch, 1); |
2342 | } |
2343 | else { |
2344 | return core->kEmptyString; |
2345 | } |
2346 | } |
2347 | |
2348 | Stringp String::AS3_charAt(double dPos) |
2349 | { |
2350 | int32_t iPos; |
2351 | INTCLAMP(iPos, dPos, this->length(), 0){ if (dPos <= -1) iPos = -1; else if (dPos >= this-> length()) iPos = this->length(); else if (!MathUtils::isNaNInline (dPos)) { iPos = (int32_t) dPos; } else { iPos = 0; } }; |
2352 | |
2353 | AvmCore* core = _core(this); |
2354 | // use unsigned compare as faster equivalent to >= 0 && < m_length |
2355 | if (uint32_t(iPos) < uint32_t(m_length)) { |
2356 | Pointers ptrs(this); |
2357 | wchar const ch = (wchar) (getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]; |
2358 | |
2359 | // newStringUTF16 does the cachedChar optimization internally, but short-circuiting the test |
2360 | // here is worthwhile because it's a pretty common case |
2361 | if (ch < 128) |
2362 | return core->cachedChars[ch]; |
2363 | return core->newStringUTF16(&ch, 1); |
2364 | } |
2365 | else { |
2366 | return core->kEmptyString; |
2367 | } |
2368 | } |
2369 | |
2370 | Stringp String::AS3_toUpperCase() |
2371 | { |
2372 | return toUpperCase(); |
2373 | } |
2374 | |
2375 | Stringp String::AS3_toLowerCase() |
2376 | { |
2377 | return toLowerCase(); |
2378 | } |
2379 | |
2380 | double String::_charCodeAtFI(int32_t iPos) |
2381 | { |
2382 | double d; |
2383 | // use unsigned compare as faster equivalent to >= 0 && < m_length |
2384 | if (uint32_t(iPos) < uint32_t(m_length)) { |
2385 | Pointers ptrs(this); |
2386 | d = (double) ((getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]); |
2387 | } |
2388 | else |
2389 | d = MathUtils::kNaN; |
2390 | return d; |
2391 | } |
2392 | |
2393 | int32_t String::_charCodeAtII(int32_t iPos) |
2394 | { |
2395 | // use unsigned compare as faster equivalent to >= 0 && < m_length |
2396 | if (uint32_t(iPos) < uint32_t(m_length)) { |
2397 | Pointers ptrs(this); |
2398 | return (getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]; |
2399 | } |
2400 | else |
2401 | return 0; |
2402 | } |
2403 | |
2404 | double String::_charCodeAtFU(uint32_t iPos) |
2405 | { |
2406 | double d; |
2407 | if (iPos < uint32_t(m_length)) { |
2408 | Pointers ptrs(this); |
2409 | d = (double) ((getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]); |
2410 | } |
2411 | else |
2412 | d = MathUtils::kNaN; |
2413 | return d; |
2414 | } |
2415 | |
2416 | int32_t String::_charCodeAtIU(uint32_t iPos) |
2417 | { |
2418 | if (iPos < uint32_t(m_length)) { |
2419 | Pointers ptrs(this); |
2420 | return (getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]; |
2421 | } |
2422 | else |
2423 | return 0; |
2424 | } |
2425 | |
2426 | double String::AS3_charCodeAt(double dPos) |
2427 | { |
2428 | int32_t iPos; |
2429 | INTCLAMP(iPos, dPos, m_length, 0){ if (dPos <= -1) iPos = -1; else if (dPos >= m_length) iPos = m_length; else if (!MathUtils::isNaNInline(dPos)) { iPos = (int32_t) dPos; } else { iPos = 0; } }; |
2430 | // use unsigned compare as faster equivalent to >= 0 && < m_length |
2431 | if (uint32_t(iPos) < uint32_t(m_length)) { |
2432 | Pointers ptrs(this); |
2433 | return (double) ((getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]); |
2434 | } |
2435 | else |
2436 | return MathUtils::kNaN; |
2437 | } |
2438 | |
2439 | int32_t String::_charCodeAtIF(double dPos) |
2440 | { |
2441 | int32_t iPos; |
2442 | INTCLAMP(iPos, dPos, m_length, 0){ if (dPos <= -1) iPos = -1; else if (dPos >= m_length) iPos = m_length; else if (!MathUtils::isNaNInline(dPos)) { iPos = (int32_t) dPos; } else { iPos = 0; } }; |
2443 | // use unsigned compare as faster equivalent to >= 0 && < m_length |
2444 | if (uint32_t(iPos) < uint32_t(m_length)) { |
2445 | Pointers ptrs(this); |
2446 | return (getWidth() == k8) ? ptrs.p8[iPos] : ptrs.p16[iPos]; |
2447 | } |
2448 | else |
2449 | return 0; |
2450 | } |
2451 | |
2452 | int32_t String::AS3_localeCompare(Atom other) |
2453 | { |
2454 | Stringp otherStr; |
2455 | |
2456 | if (AvmCore::isString(other)) |
2457 | { |
2458 | otherStr = AvmCore::atomToString(other); |
2459 | } |
2460 | else |
2461 | { |
2462 | AvmCore* core = _core(this); |
2463 | |
2464 | // legacy behavior special-cases NULL (including undefined). |
2465 | if (AvmCore::isNullOrUndefined(other)) |
2466 | { |
2467 | if (!core->currentBugCompatibility()->bugzilla585791) |
2468 | return (m_length == 0) ? 1 : 0; |
2469 | } |
2470 | |
2471 | // Proper ECMAScript behavior: |
2472 | // The 'this' object is converted to string, call it a. |
2473 | // The argument is converted to string, call it b. |
2474 | // Those strings are compared and a string comparison value is returned. |
2475 | otherStr = core->string(other); |
2476 | } |
2477 | |
2478 | return otherStr->Compare(*this); |
2479 | } |
2480 | |
2481 | #ifdef DEBUGGER |
2482 | uint64_t String::bytesUsed() const |
2483 | { |
2484 | uint64_t bytesUsed = sizeof(String); |
2485 | |
2486 | // If getType() == kDependent, the buffer's memory is account for by |
2487 | // the string upon which this string depends, so we should not include |
2488 | // it here. |
2489 | // |
2490 | // If getType() == kStatic, the buffer's memory is not on the GC heap, |
2491 | // and freeing the string would not free the buffer, so we should not |
2492 | // include its size here. |
2493 | if (getType() == kDynamic) |
2494 | bytesUsed += GC::Size(m_buffer.pv); |
2495 | |
2496 | return bytesUsed; |
2497 | } |
2498 | |
2499 | Stringp String::getMasterString() const |
2500 | { |
2501 | if (isDependent()) |
2502 | return m_extra.master; |
2503 | else |
2504 | return NULL__null; |
2505 | } |
2506 | #endif |
2507 | |
2508 | ////////////////////////////// Helpers: Width Analysis ///////////////////////////////// |
2509 | |
2510 | // The character widths structure used by the _analyze() functions below |
2511 | |
2512 | struct StringWidths |
2513 | { |
2514 | int32_t ascii, w8, w16, w32; |
2515 | }; |
2516 | |
2517 | // Analyze a UTF-8 string buffer and find out about the character widths. |
2518 | // Return false if the string contains malformed or too big sequences, |
2519 | // and strict mode is requested. |
2520 | |
2521 | // RFC 2279 |
2522 | |
2523 | // 0000 0000-0000 007F 0xxxxxxx |
2524 | // 0000 0080-0000 07FF 110xxxxx 10xxxxxx |
2525 | // 0000 0800-0000 FFFF 1110xxxx 10xxxxxx 10xxxxxx |
2526 | |
2527 | // 0001 0000-001F FFFF 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
2528 | // 0020 0000-03FF FFFF 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx |
2529 | // 0400 0000-7FFF FFFF 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx |
2530 | |
2531 | static boolbool _analyzeUtf8(const utf8_t* in, int32_t inLen, StringWidths& r, boolbool strict) |
2532 | { |
2533 | r.ascii = r.w8 = r.w16 = r.w32 = 0; |
2534 | while (inLen > 0) |
2535 | { |
2536 | uint32_t c = uint32_t (*in); |
2537 | |
2538 | switch (c >> 4) |
2539 | { |
2540 | case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7: |
2541 | r.w8++; |
2542 | r.ascii++; |
2543 | in++; |
2544 | inLen--; |
2545 | break; |
2546 | |
2547 | case 12: case 13: |
2548 | // 110xxxxx 10xxxxxx |
2549 | if (inLen < 2) { |
2550 | // Invalid |
2551 | goto invalid; |
2552 | } |
2553 | if ((in[1]&0xC0) != 0x80) { |
2554 | // Invalid |
2555 | goto invalid; |
2556 | } |
2557 | c = (((c<<6) & 0x7C0) | (in[1] & 0x3F)); |
2558 | if (c < 0x80) { |
2559 | // Overlong sequence, reject as invalid. |
2560 | goto invalid; |
2561 | } |
2562 | in += 2; |
2563 | inLen -= 2; |
2564 | if (c < 0xFF) |
2565 | r.w8++; |
2566 | else |
2567 | r.w16++; |
2568 | break; |
2569 | |
2570 | case 14: |
2571 | // 1110xxxx 10xxxxxx 10xxxxxx |
2572 | if (inLen < 3) { |
2573 | // Invalid |
2574 | goto invalid; |
2575 | } |
2576 | if ((in[1]&0xC0) != 0x80 || (in[2]&0xC0) != 0x80) { |
2577 | // Invalid |
2578 | goto invalid; |
2579 | } |
2580 | c = (((c<<12) & 0xF000) | ((in[1]<<6) & 0xFC0) | (in[2] & 0x3F)); |
2581 | if (c < 0x800) { |
2582 | // Overlong sequence, reject as invalid. |
2583 | goto invalid; |
2584 | } |
2585 | in += 3; |
2586 | inLen -= 3; |
2587 | r.w16++; |
2588 | break; |
2589 | |
2590 | case 15: |
2591 | // 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
2592 | // 111110xx ... is always invalid |
2593 | // 1111110x ... is always invalid |
2594 | // note: when 'strict' is false, we need to mimic the behavior of FP9/FP10, |
2595 | // which did not do the c&8 test. so skip it for bug-compatibility. |
2596 | if ((strict && (c & 0x08)) || (inLen < 4)) { |
2597 | // Invalid |
2598 | goto invalid; |
2599 | } |
2600 | if ((in[1]&0xC0) != 0x80 || |
2601 | (in[2]&0xC0) != 0x80 || |
2602 | (in[3]&0xC0) != 0x80) |
2603 | { |
2604 | goto invalid; |
2605 | } |
2606 | |
2607 | c = (((c<<18) & 0x1C0000) | |
2608 | ((in[1]<<12) & 0x3F000) | |
2609 | ((in[2]<<6) & 0xFC0) | |
2610 | (in[3] & 0x3F)); |
2611 | if (c < 0x10000) { |
2612 | // Overlong sequence, reject as invalid. |
2613 | goto invalid; |
2614 | } |
2615 | in += 4; |
2616 | inLen -= 4; |
2617 | r.w32++; |
2618 | break; |
2619 | |
2620 | default: |
2621 | invalid: |
2622 | // if non-strict, simply treat as character |
2623 | if (strict) |
2624 | return falsefalse; |
2625 | else |
2626 | { |
2627 | r.w8++; |
2628 | in++; |
2629 | inLen--; |
2630 | } |
2631 | } |
2632 | } |
2633 | return truetrue; |
2634 | } |
2635 | |
2636 | // Analyze a UTF-16 string buffer and find out about the character widths. |
2637 | // chars >= 0x10000 are encoded as 0xD800 + (upper 10 bits) and 0xDC00 + (lower 10 bits) |
2638 | // Returns false in strict mode if a starting surrogate pair character is not followed |
2639 | // by a correct 2nd pair character |
2640 | |
2641 | static boolbool _analyzeUtf16(const wchar* p, int32_t len, StringWidths& r, boolbool strict) |
2642 | { |
2643 | r.ascii = r.w8 = r.w16 = r.w32 = 0; |
2644 | while (len-- > 0) |
2645 | { |
2646 | wchar ch = *p++; |
2647 | if (ch <= 0x7F) |
2648 | r.ascii++; |
2649 | if (ch <= 0xFF) |
2650 | r.w8++; |
2651 | else if (ch < 0xD800 || ch > 0xDBFF) |
2652 | r.w16++; |
2653 | else |
2654 | { |
2655 | if (len == 0 || *p < 0xDC00 || *p > 0xDFFF) |
2656 | { |
2657 | if (strict) |
2658 | return falsefalse; |
2659 | // found one or two characters depending on whether the 2nd surrpgate character was present |
2660 | r.w16 += len ? 2 : 1; |
2661 | } |
2662 | else |
2663 | r.w32++; |
2664 | p++; len--; |
2665 | } |
2666 | } |
2667 | return truetrue; |
2668 | } |
2669 | |
2670 | // Create a string out of an UTF-8 buffer. |
2671 | Stringp String::createUTF8 |
2672 | (AvmCore* core, const utf8_t* buffer, int32_t len, Width desiredWidth, boolbool staticBuf, boolbool strict) |
2673 | { |
2674 | if (buffer == NULL__null) |
2675 | { |
2676 | len = 0; |
2677 | buffer = &k_zero.u8; |
2678 | staticBuf = truetrue; |
2679 | } |
2680 | if (len < 0) |
2681 | len = Length((const char*)buffer); |
2682 | |
2683 | // determine the string width to use |
2684 | StringWidths widths; |
2685 | if (!_analyzeUtf8(buffer, len, widths, strict)) |
2686 | { |
2687 | // TODO: bad character sequence error |
2688 | AvmAssert(!"bad UTF8 character sequence")do { } while (0); |
2689 | return NULL__null; |
2690 | } |
2691 | |
2692 | if (desiredWidth == kAuto) |
2693 | { |
2694 | if (widths.w32 != 0 || widths.w16 != 0) |
2695 | desiredWidth = String::k16; |
2696 | else |
2697 | desiredWidth = String::k8; |
2698 | } |
2699 | |
2700 | if (desiredWidth == String::k8 && core->kEmptyString != NULL__null) |
2701 | { |
2702 | // core has been initialized, check for cached characters |
2703 | if (len == 0) |
2704 | return core->kEmptyString; |
2705 | |
2706 | if (len == 1 && *buffer < 128) |
2707 | return core->cachedChars[*buffer]; |
2708 | } |
2709 | |
2710 | const boolbool is7bit = (widths.ascii == widths.w8) && !widths.w16 && !widths.w32; |
2711 | |
2712 | Stringp s = NULL__null; |
2713 | GC* gc = core->GetGC(); |
2714 | if (desiredWidth == k8) |
2715 | { |
2716 | if (widths.w16 != 0 || widths.w32 != 0) |
2717 | // cannot do 8-bit string with this data |
2718 | // TODO: string-too-wide error |
2719 | s = NULL__null; |
2720 | else if (staticBuf && is7bit) |
2721 | { |
2722 | // works, because we only have 7-bit ASCII |
2723 | s = createStatic(gc, buffer, widths.w8, String::k8, is7bit); |
2724 | } |
2725 | else |
2726 | { |
2727 | s = createDynamic(gc, buffer, widths.w8, String::k8, is7bit); |
2728 | uint32_t uch; |
2729 | int32_t bytesRead; |
2730 | Pointers dst(s); // NB, we assume that Utf8ToUcs4 doesn't allocate memory! |
2731 | boolbool bugzilla609416 = core->currentBugCompatibility()->bugzilla609416; |
2732 | while (len > 0) |
2733 | { |
2734 | if (*((int8_t*) buffer) > 0) |
2735 | { |
2736 | // ASCII |
2737 | *dst.p8++ = char (*buffer++); |
2738 | len--; |
2739 | } |
2740 | else |
2741 | { |
2742 | bytesRead = UnicodeUtils::Utf8ToUcs4(buffer, len, &uch, bugzilla609416); |
2743 | if (bytesRead == 0) |
2744 | { |
2745 | // invalid sequence (only if strict was false) |
2746 | *dst.p8++ = char (*buffer++); |
2747 | len--; |
2748 | } |
2749 | else |
2750 | { |
2751 | buffer += bytesRead; |
2752 | len -= bytesRead; |
2753 | *dst.p8++ = char (uch); |
2754 | } |
2755 | } |
2756 | } |
2757 | } |
2758 | } |
2759 | else |
2760 | { |
2761 | // surrogate pairs need 2 characters |
2762 | s = createDynamic(gc, NULL__null, int32AddChecked(int32AddChecked(widths.w8, widths.w16), int32AddChecked(widths.w32, widths.w32)), String::k16, is7bit); |
2763 | int32_t const convertedlen = UnicodeUtils::Utf8ToUtf16(buffer, len, s->m_buffer.p16, s->m_length, strict); |
2764 | // if this assert fires, _analyzeUtf8 didn't agree with UnicodeUtils::Utf8ToUtf16, which is bad. |
2765 | AvmAssert(convertedlen == s->m_length)do { } while (0); |
2766 | if (convertedlen < 0) |
2767 | return NULL__null; |
2768 | } |
2769 | VERIFY_7BIT(s)do { } while (0); |
2770 | return s; |
2771 | } |
2772 | |
2773 | // Create a string out of an UTF-16 buffer. |
2774 | Stringp String::createUTF16 |
2775 | (AvmCore* core, const wchar* buffer, int32_t len, Width desiredWidth, boolbool staticBuf, boolbool strict) |
2776 | { |
2777 | if (buffer == NULL__null) |
2778 | { |
2779 | len = 0; |
2780 | buffer = &k_zero.u16; |
2781 | staticBuf = truetrue; |
2782 | } |
2783 | if (len < 0) |
2784 | len = Length(buffer); |
2785 | |
2786 | boolbool is7bit = falsefalse; |
2787 | int32_t stringLength = len; |
2788 | if (desiredWidth != k16) |
2789 | { |
2790 | // determine the string width to use |
2791 | StringWidths widths; |
2792 | if (!_analyzeUtf16(buffer, len, widths, strict)) |
2793 | { |
2794 | // TODO: bad character sequence error |
2795 | return NULL__null; |
2796 | } |
2797 | |
2798 | if (desiredWidth == kAuto) |
2799 | { |
2800 | if (widths.w32 != 0 || widths.w16 != 0) |
2801 | desiredWidth = k16; |
2802 | else |
2803 | desiredWidth = k8; |
2804 | } |
2805 | else if (desiredWidth == k8 && (widths.w16 != 0 || widths.w32 != 0)) |
2806 | // cannot do 8-bit string with this data |
2807 | // TODO: string-too-wide error |
2808 | return NULL__null; |
2809 | |
2810 | is7bit = (widths.ascii == widths.w8) && !widths.w16 && !widths.w32; |
2811 | } |
2812 | |
2813 | if (desiredWidth == k8 && core->kEmptyString != NULL__null) |
2814 | { |
2815 | // core has been initialized, check for cached characters |
2816 | if (len == 0) |
2817 | return core->kEmptyString; |
2818 | |
2819 | if (len == 1 && *buffer < 128) |
2820 | return core->cachedChars[*buffer]; |
2821 | } |
2822 | |
2823 | if (desiredWidth == k16 && staticBuf) |
2824 | { |
2825 | return createStatic(core->GetGC(), buffer, len, k16, is7bit); |
2826 | } |
2827 | |
2828 | // found the width to use, now create that string |
2829 | Stringp s = createDynamic(core->GetGC(), NULL__null, stringLength, desiredWidth, is7bit); |
2830 | |
2831 | String::Pointers ptrs(s); |
2832 | if (desiredWidth == k8) |
2833 | { |
2834 | while (len-- > 0) |
2835 | *ptrs.p8++ = (char) *buffer++; |
2836 | } |
2837 | else |
2838 | { |
2839 | VMPI_memcpy::memcpy(ptrs.pv, buffer, len << desiredWidth); |
2840 | } |
2841 | VERIFY_7BIT(s)do { } while (0); |
2842 | return s; |
2843 | } |
2844 | |
2845 | PrintWriter& String::print(PrintWriter& prw) const |
2846 | { |
2847 | if (getWidth() == k8 && (m_bitsAndFlags & TSTR_7BIT_FLAG)) |
2848 | { |
2849 | // fast form |
2850 | Pointers ptrs(this); |
2851 | prw.writeN( (const char*)ptrs.p8, m_length ); |
2852 | } |
2853 | else |
2854 | { |
2855 | for (int32_t i=0, n=length(); i<n; i++) |
2856 | { |
2857 | prw << (wchar)charAt(i); |
2858 | } |
2859 | } |
2860 | return prw; |
2861 | } |
2862 | |
2863 | /////////////////////////////////////////////////////////////////////////////////// |
2864 | |
2865 | // --------------------------------------------------------------------------- |
2866 | |
2867 | // 0000 0000-0000 007F 0xxxxxxx |
2868 | // 0000 0080-0000 07FF 110xxxxx 10xxxxxx |
2869 | // 0000 0800-0000 FFFF 1110xxxx 10xxxxxx 10xxxxxx |
2870 | |
2871 | // 0001 0000-001F FFFF 11110xxx 10xxxxxx 10xxxxxx 10xxxxxx |
2872 | // 0020 0000-03FF FFFF 111110xx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx |
2873 | // 0400 0000-7FFF FFFF 1111110x 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx 10xxxxxx |
2874 | |
2875 | StUTF8String::StUTF8String(Stringp str) |
2876 | { |
2877 | if (!str || !str->length()) |
2878 | { |
2879 | m_buffer = &k_zero.c8; |
2880 | m_length = 0; |
2881 | return; |
2882 | } |
2883 | |
2884 | VERIFY_7BIT(str)do { } while (0); |
2885 | |
2886 | int32_t len = 0; |
2887 | |
2888 | MMgc::GC* gc = _gc(str); |
2889 | if (str->getWidth() == String::k8) |
2890 | { |
2891 | len = str->m_length; |
2892 | |
2893 | // if we know it's 7-bit, we don't need to walk it. |
2894 | if (!(str->m_bitsAndFlags & String::TSTR_7BIT_FLAG)) |
2895 | { |
2896 | const uint8_t* srcBuf = String::Pointers(str).p8; |
2897 | int32_t auxchars = 0; |
2898 | for (int32_t i = str->m_length; i--;) |
2899 | auxchars += (*srcBuf++ > 127); |
2900 | |
2901 | len = int32AddChecked(len, auxchars); |
2902 | |
2903 | // no hi bits? set the bit! |
2904 | if (len == str->m_length) |
2905 | str->m_bitsAndFlags |= String::TSTR_7BIT_FLAG; |
2906 | } |
2907 | |
2908 | // Deliberately using gc'ed memory here (not mmfx, unmanaged memory) |
2909 | // so that longjmp's past our dtor won't cause a long-term leak |
2910 | char* dstBuf = (char*)gc->Alloc(uint32_t(len)+1, 0); |
2911 | // can't re-use the Pointers from 7BIT above; gc->Alloc may have invalidated it |
2912 | const uint8_t* srcBuf = String::Pointers(str).p8; |
2913 | m_buffer = dstBuf; |
2914 | m_length = len; |
2915 | |
2916 | if (len == str->m_length) |
2917 | { |
2918 | VMPI_memcpy::memcpy(dstBuf, srcBuf, len); |
2919 | dstBuf[len] = 0; |
2920 | } |
2921 | else |
2922 | { |
2923 | for (int32_t i = str->m_length; i--;) |
2924 | { |
2925 | wchar ch = wchar(*srcBuf++); |
2926 | if (ch >= 128) |
2927 | { |
2928 | *dstBuf++ = char(0xC0 + ((ch >> 6) & 0x3)); |
2929 | ch = 0x80 + (ch & 0x3F); |
2930 | // fall thru |
2931 | } |
2932 | *dstBuf++ = char(ch); |
2933 | } |
2934 | AvmAssert(dstBuf - m_buffer == len)do { } while (0); |
2935 | *dstBuf = 0; |
2936 | } |
2937 | } |
2938 | else |
2939 | { |
2940 | len = UnicodeUtils::Utf16ToUtf8(String::Pointers(str).p16, str->length(), NULL__null, 0); |
2941 | if (len < 0) |
2942 | len = 0; |
2943 | char* dstBuf = (char*) gc->Alloc(uint32_t(len)+1, 0); |
2944 | m_buffer = dstBuf; |
2945 | m_length = len; |
2946 | dstBuf[len] = 0; |
2947 | // can't re-use the Pointers from Utf16ToUtf8 above; gc->Alloc may have invalidated it |
2948 | UnicodeUtils::Utf16ToUtf8(String::Pointers(str).p16, str->length(), (uint8_t*) dstBuf, len); |
2949 | } |
2950 | } |
2951 | |
2952 | StUTF8String::~StUTF8String() |
2953 | { |
2954 | if (m_buffer != &k_zero.c8) { |
2955 | // StrUTF8String is only ever stack allocated so it's OK just to free m_buffer without NULLing it first. |
2956 | GC::GetGC(m_buffer)->Free(m_buffer); |
2957 | } |
2958 | } |
2959 | |
2960 | StUTF16String::StUTF16String(Stringp str) |
2961 | { |
2962 | if (!str || !str->length()) |
2963 | { |
2964 | m_length = 0; |
2965 | m_buffer = &k_zero.u16; |
2966 | return; |
2967 | } |
2968 | |
2969 | VERIFY_7BIT(str)do { } while (0); |
2970 | |
2971 | MMgc::GC* gc = _gc(str); |
2972 | |
2973 | m_length = str->m_length; |
2974 | // Deliberately using gc'ed memory here (not mmfx, unmanaged memory) |
2975 | // so that longjmp's past our dtor won't cause a long-term leak |
2976 | wchar* dst = (wchar*) gc->Alloc(uint32ShlChecked(uint32_t(m_length)+1, String::k16), 0); |
2977 | m_buffer = dst; |
2978 | dst[m_length] = 0; |
2979 | String::Pointers ptrs(str); |
2980 | _copyBuffers(ptrs.pv, dst, m_length, str->getWidth(), String::k16); |
2981 | } |
2982 | |
2983 | StUTF16String::~StUTF16String() |
2984 | { |
2985 | if (m_buffer != &k_zero.u16) { |
2986 | // StrUTF16String is only ever stack allocated so it's OK just to free m_buffer without NULLing it first. |
2987 | GC::GetGC(m_buffer)->Free(m_buffer); |
2988 | } |
2989 | } |
2990 | |
2991 | // The following table is the length of an UTF-8 sequence, indexed by the |
2992 | // first character masked with 0x3F. It is assumed that the data is good |
2993 | // UTF-8, which is true, since we just created it :) The table is only |
2994 | // used for characters >= 0x80. |
2995 | // 0000 0080-0000 07FF 000xxxxx |
2996 | // 0000 0800-0000 FFFF 0010xxxx |
2997 | // 0001 0000-001F FFFF 00110xxx |
2998 | // 0020 0000-03FF FFFF 001110xx |
2999 | // 0400 0000-7FFF FFFF 0011110x |
3000 | |
3001 | static const uint8_t utf8Lengths[64] = |
3002 | { |
3003 | 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, |
3004 | 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2, |
3005 | 3,3,3,3,3,3,3,3,3,3,3,3,3,3,3,3, |
3006 | 4,4,4,4,4,4,4,4,5,5,5,5,6,6,6,6 |
3007 | }; |
3008 | |
3009 | StIndexableUTF8String::StIndexableUTF8String(Stringp s) : |
3010 | StUTF8String(s), |
3011 | m_lastPos(0), |
3012 | m_lastUtf8Pos(0) |
3013 | { |
3014 | m_indexable = (s->length() == this->length()); // can't init in init-list, generates warning :-( |
3015 | } |
3016 | |
3017 | int32_t StIndexableUTF8String::toUtf8Index(int32_t pos) |
3018 | { |
3019 | if (pos <= 0 || pos >= length() || m_indexable) |
3020 | return pos; |
3021 | |
3022 | // optimization: these two members kick in if this method |
3023 | // is called for positions in ascending order (which pcre usually does) |
3024 | if (pos < m_lastPos) |
3025 | m_lastPos = m_lastUtf8Pos = 0; |
3026 | |
3027 | int32_t utf8Pos = m_lastUtf8Pos; |
3028 | |
3029 | const utf8_t* p = (const utf8_t*) this->c_str() + m_lastUtf8Pos; |
3030 | for (int32_t i = m_lastPos; i < pos; i++) |
3031 | { |
3032 | utf8_t ch = *p; |
3033 | if (ch < 128) |
3034 | p++, utf8Pos++; // ASCII |
3035 | else |
3036 | { |
3037 | int32_t len = utf8Lengths[ch & 0x3F]; |
3038 | p += len; |
3039 | utf8Pos += len; |
3040 | } |
3041 | } |
3042 | m_lastPos = pos; |
3043 | m_lastUtf8Pos = utf8Pos; |
3044 | return utf8Pos; |
3045 | } |
3046 | |
3047 | int32_t StIndexableUTF8String::toIndex(int32_t utf8Pos) |
3048 | { |
3049 | if (utf8Pos <= 0 || m_indexable) |
3050 | return utf8Pos; |
3051 | |
3052 | // optimization: these two members kick in if this method |
3053 | // is called for positions in ascending order (which pcre usually does) |
3054 | if (utf8Pos < m_lastUtf8Pos) |
3055 | m_lastPos = m_lastUtf8Pos = 0; |
3056 | |
3057 | int32_t i = m_lastUtf8Pos; |
3058 | int32_t pos = m_lastPos; |
3059 | |
3060 | const utf8_t* p = (const utf8_t*) this->c_str() + m_lastUtf8Pos; |
3061 | while (i < utf8Pos) |
3062 | { |
3063 | if (i >= length()) |
3064 | break; |
3065 | utf8_t ch = *p; |
3066 | if (!(ch & 0x80)) |
3067 | i++, p++; // ASCII |
3068 | else |
3069 | { |
3070 | int32_t len = utf8Lengths[ch & 0x3F]; |
3071 | i += len; |
3072 | p += len; |
3073 | } |
3074 | pos++; |
3075 | } |
3076 | m_lastPos = pos; |
3077 | m_lastUtf8Pos = utf8Pos; |
3078 | return pos; |
3079 | } |
3080 | |
3081 | } // end namespace avmplus |